Culture and CEO Compensation

Stephen Bryan Professor of Accounting

Graduate School of Business Fordham University

New York, NY 10023 Sbryan@fordham.edu

(646) 312-8212

Robert Nash Professor of Finance School of Business

Wake Forest University Winston-Salem, NC 27109-7424

Nashrc@wfu.edu (336) 758-4166

Ajay Patel Professor and GMAC Chair in Finance

School of Business Wake Forest University

Winston-Salem, NC 27109-7424 Patela@wfu.edu (336) 758-5575

This draft: January 20, 2014 Please address correspondence to Robert Nash

We thank Kenneth Ahern, Jie Cai, Carola Frydman, Ajay Khorana, Camelia Kuhnen, Mike Lemmon, Amir Licht, Jack Rader, Raghavendra Rau, Ralph Walkling, David Yermack, Luigi Zingales and participants at the 2011 India Finance Conference at IIM-Bangalore, IESE Business School, 2012 FMA International European Meeting, 2012 FMA International Asian Meeting, 2012 FMA Annual Meeting, 2013 AFA Meeting, and the Wake Forest University Schools of Business research seminar series for helpful comments. We also thank Bill Christie for detailed comments on the paper. We are responsible for all remaining errors. Please do not quote without permission.

Culture and CEO Compensation

Abstract

This paper examines CEO compensation in 43 countries from the 1996 to 2009 period and focuses on how national culture affects contracting decisions. Our empirical analysis documents that cultural factors (measuring such societal traits as individualism and uncertainty avoidance) are significant determinants of the relative use of equity-based compensation. These factors are highly significant after controlling for the previously-identified determinants of compensation structure relating to legal environment and firm-specific characteristics. By empirically verifying the strong relation between culture and compensation, we hope to contribute to a more comprehensive understanding of the interaction between informal institutions and corporate decision-making.

1

Culture and CEO Compensation

In this study, we document that CEO compensation structure (i.e., the relative use of

equity-based compensation) differs across countries, and we seek to better understand the

determinants of this cross-sectional variation. Prior research has identified the important role of

formal institutions (e.g., law) in the development of debt and equity markets across the globe,

and therefore on the types of contracts available and held in different nations. More recent

research has considered the importance of law in determining the design of compensation

contracts across countries. This study adds to our understanding of how institutional factors

affect contracting decisions by examining whether culture also contributes to persistent cross-

sectional differences in compensation structures.

We use CEO compensation data for firms from 43 nations over the period between 1996

and 2009. We rely on dimensions of culture captured by Hofstede (2001), Schwartz (2006), and

the World Values Survey. Our study expands the application of quantitative measures of cultural

value dimensions to the financial contracting literature and provides initial evidence of how

national culture impacts executive compensation. Our empirical analysis indicates that our

measures of culture are significantly related to the cross-sectional variation in the use of equity-

based compensation. Because these results exist after controlling for previously-identified

determinants of compensation structure (i.e., firm-specific factors and legal environment), the

data strongly suggest that an understanding of culture can contribute to a better understanding of

compensation contracting.

Additionally, we find anecdotal evidence in our data indicating a potential relation

between culture and compensation structure. As one example, the Delhaize Group (a Belgian

company with U.S. operations) explicitly states in its annual report that it incorporates

2

differential pay practices across executives in different countries based upon regional norms, and

it notes that equity-based compensation is more widely prescribed in the U.S. than in Europe.

Further, as we see in the below quotation from its 2008 annual report, Delhaize encourages the

following ownership levels based upon country-specific norms:

“. . . [D]uring their active employment, each of our Chief Executive Officers and our other executive officers is expected to acquire and maintain ownership of Delhaize Group stock equal to a multiple of such officer’s annual base salary. These multiples are as follows: Chief Executive Officer: 300% of annual base salary; U.S.-based executive officers: 200% of annual base salary; European-based executive officers: 100% of annual base salary.”

Although our focus is on differences across firms in different countries (rather than differences

within firms, as the Delhaize example illustrates), the Delhaize disclosure lends additional

credence to operationalized differences in compensation structure, explicitly attributed to

informal factors, including aspects of cultural norms.

Thomas Friedman (2005) has famously argued that the forces of globalization are

mitigating the effects of cross-national differences so that the world is becoming “flat”. If

applied within the context of corporate governance, this “flat world” perspective would be

consistent with the convergence hypothesis articulated by Coffee (1999). That is, the most

effective compensation structures will become more widely-adopted and cross-country

differences in compensation design should dissipate over time.

However, recent empirical evidence is inconclusive regarding such a convergence (or

“flattening” of the world) with respect to executive compensation. Fernandes et al. (2012) test for

convergence in various compensation policies across a sample of 14 countries from 2003-2008.

After controlling for firm-level and country-level effects, they provide evidence of compensation

convergence for their sample countries during their sample period. Alternatively, after

considering a longer time series, Bryan et al. (2010, 2011) indicate that there is no recent

3

international convergence towards U.S. standards of compensation structure. These Bryan et al.

studies primarily attribute the cross-sectional variance in CEO compensation structure to

differences in the legal environment. However, in addition to the legal systems, other types of

institutional variables may impact compensation structure. Dyck and Zingales (2004) and Stulz

and Williamson (2003) identify that financial decisions may be affected by “extra-legal”

characteristics (such as social or cultural norms). While the intersection of culture and economics

has attracted increasing attention in the finance literature (e.g., Ahern, Daminelli, and Fracassi

(2012), Breuer and Salzman (2010), Giannetti and Yafeh (2012), and Guiso, Sapienza, and

Zingales (2006, 2009)), the impact of culture on compensation design has largely been

unexplored.

We initially draw from the work of two Nobel Prize winning economists to consider the

impact of culture on compensation design. Coase (1937) explains that “institutions matter”;

North (1990) documents that institutions are the “rules of the game in a society” that “consist of

both informal constraints and formal rules”. While the contracting implications of the “formal

rules” (such as the legal and regulatory environments) have been more thoroughly examined in

the law and finance and in the regulatory economics literatures, the role of the “informal

constraints” has drawn much less attention from financial economists. Therefore, we focus on

these informal constraints in our efforts to contribute to a more comprehensive understanding of

the interaction between institutions and corporate decision-making.

Indeed, we find that culture matters. We first confirm the persistence of significant

differences in the design of compensation contracts across countries by studying CEO

compensation structures for firms in 43 nations. We then focus on the source of this cross-

sectional variation. While we support the contentions of the “law and finance” literature by

4

verifying that some of the cross-sectional variation is related to differences in legal environment,

our data also indicate a strong role for culture in explaining the design of compensation

contracts. Specifically, we draw from the sociology and cross-cultural psychology literatures to

identify cultural factors that theory predicts should influence contracting decisions. Our

empirical analysis confirms that cultural factors (measuring such societal traits as individualism

and uncertainty avoidance) are significant determinants of the relative use of equity-based

compensation.

We recognize that concerns may be raised about our study. First, endogeneity may be

problematic when compensation structure is explained using variables that are jointly

determined. For purposes of this study, culture is deemed to be predetermined, instead of jointly

determined, with regard to observed values of compensation structure in our sample. Second, we

recognize that our findings may be driven by an omitted variable that is correlated with culture

and with CEO compensation structure. We attempt to address this concern by rerunning our

empirical models after randomly assigning values to the cultural factors studied. We then retest

the significance of our findings based on the simulated distribution of slope coefficients. We find

our results are not driven by chance and that the cultural factors we investigate are significantly

related to the relative use of equity-based compensation.

To further investigate whether culture matters, we test the significance of the factors in

two subsets of the data. We separate firms that produce tradable goods (i.e., products that may be

sold internationally) from those with non-tradable goods (i.e., products that may be sold only

domestically). Since firms primarily selling non-tradable goods should be domestically-focused

and should be more likely to hire CEOs locally, culture should play a more important role in

explaining compensation structure for firms with local markets (non-tradable goods), as opposed

5

to global markets. Our findings are consistent with this hypothesis. Cultural factors are more

significant for firms that produce non-tradable goods. However, cultural factors continue to play

a role, albeit smaller, in explaining compensation structure for firms with tradable goods. We

feel that these additional tests help reduce concern about an omitted variables problem. We

describe these tests in greater detail later in the paper.

Our findings are robust to the use of alternative instruments identified in the literature to

measure cultural dimensions. Furthermore, our results are robust to different partitionings of the

data and are also robust to the inclusion of additional institutional variables (such as measures of

the religion, language, and the tax environment).

We organize the remainder of the paper as follows. Section I describes the intersection of

culture and contracting decisions and introduces our cultural variables. In Section II, we review

the previously-established factors affecting the structure of compensation contracts. Section III

documents our data sources and sample and provides descriptive statistics. We present our

primary empirical results in Section IV and conduct robustness testing in Sections V and VI.

Section VII provides a summary and conclusion.

I. The Role of Culture in Contracting Decisions

A. Measures of Culture

Drawing from North’s (1990) analogy of institutions as defining “the rules of the game”,

Siegel et al. (2011) refer to culture as “the unwritten, unspoken rules of the game”. While

perhaps unwritten and unspoken, culture is not unmeasurable, at least according to the cross-

cultural psychologists. The field of cross-cultural psychology has specified numerous paradigms

for identifying international cultural differences. These quantifiable measures of societal traits

6

allow for the formulation of testable hypotheses regarding culture’s impact on economic

decision-making. As summarized by Sondergaard (1994), one of the most widely-utilized and

empirically-validated typologies of culture was developed by Geert Hofstede (1980).

Hofstede (1980) defines culture as the “software of the mind”. Hofstede uses values as a

means to operationalize his dimensions of cultural variation (known as cultural value dimensions

or CVD).1 Hofstede’s cross-cultural study to identify and quantify values is rooted in

psychology and sociology theory and focuses on the fundamental challenges encountered by all

societies. Hofstede (1980) concludes that cultures face four basic problems: 1) the relationships

between the individual and the group, 2) the inequitable distribution of power, 3) the social

implications of gender, and 4) the ability to tolerate uncertainty. Theory should also help to

delineate the expected societal response or the expected means of addressing these pervasive

cultural dilemmas. Hofstede argues that each of his value dimensions reflects a particular

culture’s solution to each societal problem. Specifically, Hofstede’s cultural value dimensions

provide a bi-polar spectrum of possible responses as a means of coping with each of the four

major issues.

Hofstede’s model identifies where a country’s culture ranks along a spectrum for each of

the four issues. He uses detailed interviews to gauge the society’s stance regarding the various

values and identifies quantifiable measures that discriminate among cultures through factor

analysis and other statistical techniques. Hofstede’s original study involved over 100,000 surveys

conducted in 50 countries. Sondergaard (1994) reports that Hofstede’s work has been

subsequently expanded (covering 74 countries and regions), has been widely-cited (over 6,225

citations per SCCI for 1980-2011), and has been broadly validated through replication.

1Siegel et al. (2011) further support the use of CVD by noting that cultural values are the standards that outline appropriate behavior in various circumstances and are paramount in establishing frameworks for people vis-à-vis their relationships with each other.

7

Accordingly, we use Hofstede’s cultural value dimensions (CVD) in our empirical analysis

(although we employ several other of the major typologies of culture as part of our robustness

testing).2

B. Cultural Dimensions and Agency Costs

Jensen and Meckling (1976) argue that firms may use equity-based compensation to

better align the interests of managers and stockholders. However, the potential severity of these

conflicts of interest may differ across firms and also across countries. To better understand the

country-level influences, we draw from Hofstede’s (1980) measures of cultural value

dimensions. We focus on the Hofstede cultural value dimensions that should most directly affect

agency conflicts, and thus should most directly affect the structure of executive compensation.

Of the four dimensions specified by Hofstede, our analysis of the literature (described below)

suggests that Hofstede’s measures of Individualism and Uncertainty Avoidance are the most

likely to directly affect the potential severity of the types of agency conflicts most likely to be

addressed through compensation contracts.3

1. Individualism

Agency problems of equity occur when managers pursue individual self-interests at the

expense of shareholders. The Hofstede (1980) measure of Individualism (vs. Collectivism)

appears especially relevant to this potential conflict of interest. Hofstede (1991) describes that a

society’s Individualism dimension reflects the relation between the individual and the group and

demonstrates the degree to which individual interests may prevail over collective interests. That

2 For example, Schwartz (2006) provides a similar typology based on bi-polar dimensions or “value types”. We use cultural measures from Schwartz and from other researchers in our robustness analysis. 3Hofstede has specified additional cultural value dimensions. However, as we describe in the following sections, we feel that Individualism and Uncertainty Avoidance are the most theoretically justifiable cultural factors to consider when examining the role of compensation structure in aligning incentives. Accordingly, we do not think that it is appropriate or relevant to include other Hofstede variables in our analysis of compensation structure.

8

is, members of a society ranking high on the Individualism spectrum are more inclined to pursue

their own self-interest and be less attuned to the interests of the group. Individualistic societies

involve loosely-knit social relations, emphasize self-reliance and independent action, and

embrace personal challenge and individual freedom. Conversely, societies scoring lower on

Hofstede’s Individualism scale are characterized by more tightly-connected and cohesive in-

groups, where the general interests of the collective prevail over those of the individual.

Accordingly, in a more individualistic society, the interests of the manager (i.e., the individual)

should be less closely aligned with the collective (i.e., the shareholders or the “group” of

minority owners). As a result, the agency costs of equity should be positively related to the

Individualism measure. This leads to our first hypothesis regarding the impact of culture on

compensation structure. Since equity-based compensation helps to reduce stockholder/manager

conflicts, we expect a greater relative use of equity-based compensation in countries with

cultures scoring high on Individualism.

2. Uncertainty Avoidance

A society’s level of comfort with uncertainty (or its level of Uncertainty Avoidance, to

use Hofstede’s nomenclature) is another cultural characteristic that may affect compensation

structure. The Hofstede (1980) Uncertainty Avoidance index (UAI) measures the degree to

which people feel stress or discomfort in unstructured or risky circumstances. This metric takes

on higher values in cultures that crave greater predictability of outcomes and perceive ambiguity

as a threat. Societies scoring high on the UAI spectrum seek greater stability and are less

tolerant of conflict and competition (especially when outcomes are indeterminate). Conversely,

societies that Hofstede characterizes as having low degrees of UAI are more tolerant of

ambiguity, are more willing to take risks, and are more secure in unstructured situations.

9

Cultures with lower UAI are also more accepting of competition and more comfortable with

conflict and confrontation.

Several empirical studies investigate the economic implications of a society’s propensity

for uncertainty avoidance. These studies support Hofstede’s theoretical prediction that entities in

societies with higher levels of UAI typically make decisions that reflect a greater degree of risk-

aversion. For example, Griffin et al. (2008) report a negative relation between Uncertainty

Avoidance and corporate risk-taking. Kogut and Singh (1988) find that firms from cultures

exhibiting higher Uncertainty Avoidance choose more conservative entry methods when

engaging in foreign direct investment. Ramirez and Tadesse (2007) document that firms from

cultures with higher Uncertainty Avoidance hold larger cash balances. Kirkman, Lowe, and

Gibson (2006) provide an excellent summary of other studies that further demonstrate the

empirical validity of Hofstede’s UAI.

Regarding the specific impact of UAI on compensation structure, we expect a negative

relation between a nation’s UAI and the use of equity-based executive compensation.

Specifically, if high UAI cultures favor stability and security and strongly seek to avoid

uncertainty, CEO compensation contracts of firms in high UAI countries should involve a greater

use of cash-based compensation and a lower use of the riskier, equity-based compensation.

Alternatively, if the culture is more tolerant of risk-taking and more willing to accept ambiguity

of outcome, CEO compensation design should include a greater amount of equity-based pay.

3. Alternative Instruments of Culture

We also recognize that the sociology and cross-cultural psychology literatures provide a

rich array of metrics for expressing cultural differences. To help confirm the robustness of our

results, we identify alternative variables that serve as instruments for cultural dimensions.

10

3a. Schwartz Variable: Mastery vs. Harmony

We complement our use of the Hofstede dimensions by considering the cultural

orientation scores developed by Schwartz (2006). Based on extensive interviews conducted

during 1988-2004, Schwartz derived three broad measures of societal traits, with each following

a bi-polar spectrum. In many respects, the Schwartz spectra are similar to Hofstede’s cultural

value dimensions. However, the Schwartz variable Mastery (vs. Harmony) appears to capture

additional cultural attributes (not explicitly measured by Hofstede) that may also affect the

agency conflicts that compensation structure may be designed to mitigate. Specifically, the

Mastery/Harmony variable conceptualizes the degree of a society’s acceptance of entrepreneurial

endeavors and individual assertiveness.

A firm in a high-Mastery culture may be more adversely affected by the agency problems

of equity. We expect that managers may be more likely to undertake actions that provide

personal benefits (at the possible expense of the best interests of shareholders) if the national

culture approves of and attaches higher social value to personal success and active self-

assertion.4 Accordingly, since equity-based compensation is one means to better align

stockholder/manager interests, we expect a greater use of equity-based compensation in cultures

scoring high in Mastery (and expect a lower use of equity-based compensation in cultures

scoring high in Harmony).

3b. Language

4 Chui et al. (2002) also consider the relation between scores on the Schwartz (2006) Mastery/Harmony spectrum and agency costs of equity. Specifically, they contend that a higher Mastery score suggests a greater propensity for stockholder/manager conflicts (especially the “free cashflow problem”). Since Jensen (1986) notes that the use of debt helps to mitigate the “free cashflow problem”, Chui et al. (2002) expect that firms from cultures with high Mastery scores should use more leverage (to better align the interests of the managers and stockholders). The study confirms a positive relation between Mastery scores and the level of leverage in capital structures. We perform similar tests, but focus on compensation structure as the primary method to address agency conflicts.

11

In seminal papers linking the culture and “law and finance” literatures, Stulz and

Williamson (2003) and Licht et al. (2001) use language as an instrument for culture. Language

appears to be a good instrument for culture, since Siegel et al. (2011) consider it an exogenous

factor and Hallpike (1986) documents that people sharing a language also share many

fundamental societal principles and world-views.5 Stulz and Williamson (2003) and Licht et al.

(2001), focusing on the connection between language and legal environments across countries,

note that legal protection of shareholder rights is stronger in English common-law countries.

Stulz and Williamson (2003) and Licht et al. (2001) raise the issue that these differences may

have a cultural genesis. Specifically, Licht et al. (2001) suggest that the “cultural infrastructure”

of the English-speaking countries may affect the differences in the legal systems between the

common-law and the other nations. Accordingly, our robustness tests include language as an

alternative means of assessing the interaction of culture and compensation structure.

3c. Religion

Guiso et al. (2003) focus on religion as a fundamental determinant of various differences

in societal values and economic decision-making. These authors consider how religion affects

trust (Guiso et al., 2003), bilateral trade (Guiso et al., 2006), and stock market participation

(Guiso et al., 2008). Stulz and Williamson (2003) also identify religion as a proxy for culture

and demonstrate that religion has significant explanatory power in understanding variation in

legal systems across countries.

Additionally, religion may underlie aspects of our primary measures of culture. We

recognize a potential connection between religion and Hofstede’s (1980) UAI. Hofstede (1983)

observes that cultures scoring high on Uncertainty Avoidance feel greater stress from ambiguity

5Guiso et al. (2009) further support the connection between language and culture by identifying that a common language is one of the most significant determinants of trust between nations.

12

and seek to minimize indeterminate outcomes. He further contends that, particularly in societies

characterized by high levels of Uncertainty Avoidance, religion may serve as a coping

mechanism by helping to make uncertainty more tolerable.6 Hofstede (1983) argues that cultures

high in Uncertainty Avoidance seek institutions providing order and predictability and gravitate

towards religions that are more rigidly structured and are more likely to claim absolute truth.

Such religions reduce ambiguity and make uncertainty more bearable (which is especially

important to people highly stressed by uncertainty). Furthermore, LaPorta et al. (1997b) note that

hierarchical religions (such as Catholicism) emphasize strict, vertical bonds of authority that may

help to instill the order and structure favored in a culture ranking high in Uncertainty Avoidance.

If countries with more hierarchical religions (e.g., Catholicism) are generally

characterized by greater Uncertainty Avoidance, we would expect that firms in countries with

hierarchical religions should use less equity-based compensation. We follow Kwok and Tadesse

(2006) and use religion as an instrument for a culture’s tolerance for uncertainty (Uncertainty

Avoidance). Therefore, since prior research suggests a positive relation between Catholicism

and Hofstede’s measure of Uncertainty Avoidance, we repeat our empirical analysis using

religion as an alternative measure of culture.

3d. Trust

While the Hofstede and (to a lesser extent) Schwartz variables have dominated the

empirical literature examining the consequences of culture, Sapienza et al. (2010) note that trust

is another cultural factor that is becoming more widely studied. Sapienza et al. (2010) report that

over 500 papers have involved measures of trust (primarily derived from the World Values

6Iannaconne (1998) and Hull and Boyd (1989) also identify a potential relation between religion and a society’s tolerance of uncertainty. These authors argue that many standard elements of religious institutions were developed to mitigate risk.

13

Survey).7 Therefore, since measures of trust may provide alternative means of assessing cultural

differences, we also include a trust variable in our robustness tests. We follow Aghion et al.

(2010) and draw our empirical proxy from the World Values Survey (WVS) item focusing on

trust of business. The specific WVS question gauges the individual’s degree of confidence or

trust in companies. We focus on the distrust of companies because it is a lack of trust within the

firm that contributes to the agency conflicts that compensation structure is designed to address.8

We expect that firms from societies reflecting lower levels of trust should use less equity-

based compensation. Specifically, Guiso et al. (2008) find that individuals in cultures exhibiting

lower levels of trust hold less of their total wealth in equities. Since our measure of

compensation structure reflects the proportion of total compensation provided by equity, we

similarly contend that a lower level of trust will be associated with a lower use of equity-based

compensation. Guiso et al. (2008) also note that national stock markets are less developed in low

trust cultures. Bryan et al. (2010) identify a lower use of equity-based compensation in countries

with less developed equity markets.9 Accordingly, if lower trust contributes to a less developed

stock market, we should expect a lower use of equity-based compensation in nations where trust

of business is low.

II. Law, Agency Costs, and the Structure of Compensation Contracts

7The extant literature has demonstrated a strong connection between trust and various economic outcomes. For example, recent empirical studies have shown that a society’s lack of trust can be linked to a diminished level of overall economic growth (LaPorta et al., 1997b), a lower degree of stock market participation (Guiso et al., 2008), and a reduced amount of bilateral trade and foreign direct investment (Guiso et al., 2009). 8Fukuyama (1995) notes that large organizations, such as those in our sample, require cooperation between strangers. Such cooperation is heavily predicated on trust. Similarly, Coffee (2001) refers to trust as the critical “organizational glue”. The level of trust within the culture affects the strength of this “glue” and the potential severity of agency conflicts. 9 Stock prices must be informationally efficient to provide an accurate retrospective of firm performance and a reasonable means of motivating and compensating managers. If the stock market is not informationally efficient, share price has diminished value for monitoring and for linking pay to performance.

14

A. The Legal Environment and Compensation Structure

While we focus on the potential impact of culture on compensation structure, we

recognize that other institutional characteristics (primarily the nation’s legal system) may also

affect contract design. Bryan et al. (2010, 2011) provide empirical evidence that differences in

legal systems contribute to the cross-sectional variation in compensation structures. Most

commercial law derives from one of two broad traditions: common law or civil law. Common

law is determined primarily by judges in legal systems where laws are formulated through

precedent and then later incorporated into the legislature. Common law is based on English

tradition and, like the other systems, has been spread across the world mostly through occupation

and colonization. Civil law, on the other hand, relies on statutes and comprehensive codes,

formulated and articulated primarily by legal scholars and governmental authorities.

One of the more important areas of difference among legal systems is the level of

protection afforded to stockholders and the relative strength of shareholder rights. We use the

Anti Self-Dealing Index (developed by Djankov et al. (2008)) to measure the degree of legal

protection for shareholders. The Anti Self-Dealing Index focuses on shareholder protection

against expropriation by insiders. The index explicitly considers such control mechanisms as

legally mandated disclosure and approval processes, as well as procedural remedies and penalties

if minority shareholders are wronged. As described by Djankov et al. (2008) and Johnson,

LaPorta, Lopez-de-Silanes, and Shleifer (2000), insider self-dealing (i.e., expropriation by

majority owners or “tunneling”) is a major concern to minority shareholders. The Anti Self-

Dealing Index directly targets the level of legal protection against this risk and has been used

extensively in recent studies of the intersection of law and finance. Furthermore, the quality of

each nation’s law enforcement may also affect contracting decisions. In other words, the legal

15

rights of investors can be undermined if a country has ineffective enforcement of well-designed

laws. We measure the effectiveness of law enforcement in each country using the Rule of Law

index from LaPorta et al. (1997a). This variable takes on a higher value for nations with a

stronger tradition of law and order.

We expect a greater use of equity-based compensation for CEOs in countries providing

stronger protection of shareholder rights. This is primarily because equity-based compensation

will be most effective when stock prices are informationally efficient. Bryan et al. (2010, 2011)

provide evidence consistent with this hypothesis.

B. Firm-Specific Agency Costs and Compensation Structure

Most studies of the determinants of U.S. compensation structure have focused on the role

of agency costs of debt and equity. When assessing the potential severity of conflicts within a

firm, the contracting literature primarily examines enterprise-specific characteristics (e.g., the

firm’s growth options, size, free cash flow, and leverage).

Authors such as Bryan et al. (2000), Kole (1997), Bizjak, Brickley, and Coles (1993), and

Gaver and Gaver (1993) contend that firms with greater amounts of growth options have broader

informational asymmetries that create a larger potential for opportunistic behavior by managers.

As a result, these firms should use more equity-based compensation. Our proxy for the

prevalence of growth options is the ratio of the market value to the book value of the firm’s

assets (MVBV). This ratio is equal to the book value of assets minus the book value of equity

plus the market value of equity divided by the book value of assets.

Jensen and Meckling (1976) contend that agency costs increase with firm size since a

larger span of operations allows for greater managerial opportunism and contributes to less

effective external monitoring. Previous studies (e.g., Yermack (1995)) have found that bigger

16

firms pay managers with significantly larger relative amounts of equity-linked compensation.

These studies also attribute this relation to the greater degree of difficulty in monitoring

managers of larger companies. As used by Gabaix and Landier (2008) and Baker and Hall

(2004), our proxy for size is the natural logarithm of the firm’s total assets. We predict a positive

relation between firm size and the relative use of equity-based compensation.

Yermack (1995) and Bryan et al. (2000) have shown that excess cash is negatively

related to equity-based pay. We measure the firm’s free cash flow by using Lehn and Poulsen’s

(1989) cash flow statistic. This proxy for the firm’s free cash flow is: operating income before

depreciation less income tax, interest, and dividends paid. This measure of free cash flow is

divided by the sum of the firm’s market value of equity and book value of debt. The FCF ratio

provides an indication of cash availability.

Finally, prior research has also shown a negative relation between a firm’s use of debt

and the use of equity in the compensation structure. Specifically, John and John (1993), Yermack

(1995), and Bryan et al. (2006) contend that higher levels of leverage exacerbate agency conflicts

between stockholders and bondholders, increasing the agency cost of debt. Equity-based

compensation would work against alleviating this conflict within a highly-levered firm. We

investigate this issue by examining whether the ratio of a firm’s book value of total debt to total

assets is related to the use of equity-based compensation.

III. Data and Descriptive Statistics

The following section describes the data that we use to measure and explain patterns in

the design of compensation contracts. To obtain detailed compensation information for a wide,

cross-section of countries, we examine the financial disclosures of 589 firms (from 42 countries)

17

that issued ADRs during 1996-2009.10 As we will describe below, these ADR issuers must

provide documentation of compensation structure in their filings with the SEC. Therefore,

examining ADR issuers is an effective way to acquire executive compensation data covering a

wide sample of non-U.S. firms for a long time period.11 We also examine detailed compensation

information for all U.S. firms from the ExecuComp database.

A. Data Sources

The ability of researchers to investigate cross-country differences in compensation policy

has been limited by a lack of a long time series of readily available, detailed, and consistently

presented compensation data for a broad array of countries. The key to our study is the

information made available through the SEC’s Form 20-F. Form 20-F is a reporting requirement

for foreign firms whose equity trades in the U.S. market through American Depository Receipts

(ADRs). Sponsored ADRs (those managed by a depository bank) are classified as Level 1,

Level 2, Level 3, or Rule 144A. Level 1 ADRs are traded over-the-counter, while Rule 144A

ADRs are private placements. Both are exempt from a majority of U.S. reporting requirements

(such as filing the Form 20-F). Level 2 ADRs are listed on an exchange, or quoted on NASDAQ.

Level 3 ADRs are for new equity offerings. Level 2 and Level 3 ADR issuers must file the Form

10 Our use of ADR-issuers as our sample of non-U.S. firms biases against a finding that cultural differences affect contract design. By definition, ADR-issuers have chosen to pursue a cross-national ownership base and have taken steps to establish a presence in the U.S. financial markets. 11 As in most empirical work, there are tradeoffs involved in the selection of data sources. For example, Desai, Foley, and Hines (2004), Booth et al. (2001), Rajan and Zingales (1995), and Megginson, Nash, and Van Randenborgh (1994) note that using cross-country data may lead to measurement error when comparing countries (due to inconsistencies in accounting practices). However, SEC filings must follow either U.S. generally accepted accounting principles or International Financial Reporting Standards. U.S. firms must follow the former and non-U.S. firms have a choice of either. Differences in these sets of standards have narrowed over the years, and standard-setters have the goal of full convergence. Moreover, our data source for firm-specific financial information is Compustat, which standardizes the data even further. Other, more recent sources of cross-national data (such as Capital IQ) present a much more limited time series of data. Also, by including data from non-ADR issuers, Capital IQ can provide fewer assurances regarding the standardization of accounting information.

18

20-F and thus provide the compensation data that we use in our analysis.

In compiling our sample, we first search the Compustat database for a listing of all

companies with Level 2 and Level 3 ADR programs during 1996-2009. Next, we use the

Securities and Exchange Commission’s EDGAR database to hand-collect information from the

Form 20-F filings. We obtain information to measure the value of stock options, restricted stock,

and cash-based compensation to analyze cross-country differences in the design of the firm’s

compensation structure. We also gather information about any other forms of managerial

compensation (such as non-equity payouts from long-term incentive plans) to include in our

analysis of compensation structure.

B. Primary Components of Compensation Structure

We measure the value of options granted using the Black-Scholes (1973) model. We

obtain country-specific risk-free rates from the IMF’s International Financial Statistics

Yearbook. We calculate dividend yields from Compustat data and stock return volatility from

the Center for Research in Securities Prices (CRSP) database. As is standard in the

compensation literature (Yermack, 1995; Bryan et al., 2000), the option expected time to

maturity is 10 years for all firms, unless stated otherwise.

Our primary measure of compensation structure focuses on the relative use of equity-

based pay. Equity-based compensation is the sum of the value of stock options and restricted

stock. We define the relative use of equity-based compensation (EquityPct) as the value of

equity-based compensation divided by total compensation, where total compensation is the sum

of equity-based compensation, long-term incentive plans, and cash compensation.

To more thoroughly examine the relation between the compensation structures of U.S.

and non-U.S. companies, we complement our ADR sample with data for U.S. firms. For the

19

U.S. sample, we obtain information on all components of CEO compensation, as defined above,

from the Standard & Poor’s (S&P) ExecuComp database. ExecuComp provides detailed data for

companies included in the S&P 500, S&P 400 MidCap, and S&P 600 SmallCap indexes.

We use Compustat to form the enterprise-level control variables in our empirical models

for both U.S. firms and our ADR sample. Our final sample covers the period 1996-2009 and

consists of 2,801 firm-year observations from 42 non-U.S. countries and 24,644 firm-year

observations for U.S. firms.

C. Primary Measures of National Culture

We measure national culture primarily through Hofstede’s dimensions of Individualism

and Uncertainty Avoidance. As we described earlier, Hofstede’s (1980) cultural value metrics

provide a very widely-cited and rigorously validated typology of cross-country differences in

culture. We obtain the most recent values of the Individualism and Uncertainty Avoidance

variables from Hofstede’s website.12

D. Descriptive Statistics

Table I indicates that there is clustering in our data with respect to the number of firm-

year observations across countries, in addition to variation across time within each country. The

ADR-issuing firms in our sample are more likely to be geographically close to the U.S. (e.g.,

Mexico, Brazil, or Chile), to have strong trade relations with the U.S. (e.g., France, Germany,

Italy, Japan, and China), or to have capital-market orientations similar to the U.S. (e.g.,

England).13 These findings are consistent with the evidence in Bryan et al. (2010), Pagano,

Roell, and Zechner (2002), and Tesar and Werner (1995).

12 Detailed information about Hofstede’s work as well as data for each of his value dimensions are available at www.geert-hofstede.com. 13 Again, each of these similarities biases against a finding that cultural differences are related to cross-sectional

20

Insert Table 1 about here

Figure 1 presents the time-series variation in the use of equity in the CEO compensation

mix for U.S. and non-U.S. firms. For each year during the 1996-2009 period, non-U.S. firms

used substantially less equity-based compensation than U.S. firms. Despite this consistent cross-

sectional difference between U.S. and other companies, the pattern of changes in the equity-

based pay of U.S. companies over time appears to be largely mirrored by changes in the

compensation structures of companies in other countries.

Insert Figure 1 about here

As shown in the figure, equity-based pay as a percentage of total CEO compensation for

U.S. companies increased from about 35% in 1996 to just under 50% in 2001. It then declined to

approximately 42% in 2003, remaining at that level through 2005, before falling to 35% in 2006.

However, it had climbed back to around 48% by 2008 before moving to about 42% in 2009.

For non-U.S. firms, by comparison, equity represented less than 5% of the CEO

compensation mix in 1996, climbing to just over 20% in 2001 before declining to approximately

12% in 2002. Equity-based compensation stayed at around 12% through 2004, declined to 8%

by 2008, and rose to around 14% in 2009.

The important point from Figure 1 is that not only did U.S. firms use more equity-based

compensation than non-U.S. firms, the cross-sectional difference in CEO compensation structure

between those two groups of firms did not narrow over the 14-year sample period. This finding

is consistent with our hypothesis that CEO compensation structures vary across nations due to

variation in compensation structure.

21

country-level differences, and is inconsistent with the notion that globalization of labor and

capital markets has led to a convergence in CEO pay practices.

Table II provides information on the average use of equity in the CEO’s compensation

structure by firms from each of the countries in our sample. The use of equity-based

compensation is higher for countries that have an English Common Law origin than for those

with non-English Common Law origin. For example, firms from countries such as Australia

(33.4%), Bermuda (44.3%), England (25.6%), and the U.S. (43%) use more equity in the CEO’s

compensation mix than those from France (15.5%), Germany (16.3%), and Japan (3.8%).

Insert Table II about here

Table II also indicates that the non-U.S. firms issuing Level 2 and Level 3 ADRs tend to

be large relative to the average U.S. firm. If these large firms are issuing ADRs to tap into the

deep U.S. capital markets and are willing to bond themselves to U.S. oversight and governance,

they are likely to be more similar to U.S. firms than their local counterparts.14 If true, this should

bias our results against finding significant differences in compensation structure across countries.

More important, if these non-U.S. firms behave like those in the U.S., differences in national

culture should not help explain any observed differences in compensation structure. To the

extent we determine that culture matters, the findings should be even more significant in a

sample that includes local firms that did not issue ADRs.

Figure 2 presents the time series variation in the use of equity-based compensation by

partitioning the sample by legal origin – firms from countries with English Common Law origin

versus those with non-English Common Law origin (i.e., civil law systems). To ensure that our

14 For detailed analysis of ADR-issuers, see Lang et al. (2003), Pagano et al. (2002), Reese and Weisbach (2002), among many other studies of the cross-listing process.

22

findings are not driven by the pay practices of U.S. firms, we further partition the data for firms

from countries with English Common Law origin into two groups: U.S. and non-U.S. firms.

Insert Figure 2 about here

Figure 2 indicates that firms from countries with English Common Law origin use more

equity in the CEO’s compensation structure than those from countries with non-English

Common Law origins. Moreover, within the common law family, U.S. firms use larger amounts

of equity-based compensation than firms from non-U.S. English Common Law countries. The

cross-sectional differences between these three groups of firms do not narrow over time. This

suggests that factors other than legal origin must contribute to those consistent differences.

We next investigate whether cultural differences are associated with the cross-sectional

variation in compensation structure. We initially focus on Hofstede’s measures of Individualism

and Uncertainty Avoidance as measures of national culture that should relate to the cross-

sectional variation in compensation structure. We hypothesize that firms from countries scoring

higher on Hofstede’s measure of Individualism should use larger levels of equity in their

compensation structure. Similarly, firms from countries characterized by higher levels of UAI

should use less equity, and more cash, in the compensation structure.

To examine whether compensation structures are related to cultural differences, we first

divide the full sample of firms into quartiles based on country-level values for Individualism and

UAI.15 Figure 3 graphs the average equity structure by quartile for our total sample (i.e., that

includes U.S. firms), where we form quartiles based on country-level values for Individualism.

15Since the quartiles are based on country-level values of the two cultural variables, the sample contains approximately equal numbers of countries within each quartile. However, since our compensation data are at the level of the firm, each quartile contains different numbers of firm.

23

Consistent with our hypothesis, firms from cultures with higher levels of Individualism use more

equity in the compensation mix.

Insert Figure 3 about here

Figure 4 presents the average compensation structure for firms in each quartile based on

Hofstede’s UAI metric. As expected, firms from countries characterized by high levels of UAI

use lower amounts of equity-based compensation.

Insert Figure 4 about here

To ensure that the results are not being driven by the U.S. observations in our sample, we

also create quartiles using only the sample of non-U.S. firms. Figures 5 and 6 provide average

compensation structures for firms in each quartile based on the country-level values of

Individualism and UAI. The results indicate that for non-U.S. firms, the average use of equity-

based compensation is positively related to the country’s level of Individualism and negatively

related to the country’s level of UAI.

Insert Figures 5 and 6 about here

Table III provides descriptive statistics for compensation structure and for firm-specific

control variables (size, market-to-book ratio, leverage, and free cash flow) for quartiles formed

from the legal and the cultural determinants that we investigate in this paper. The legal variables

are the Anti Self-Dealing Index and the Rule of Law. The cultural variables are Individualism and

UAI.

Insert Table III about here

24

As we report in Table III, the data indicate that (for quartiles based on the Anti Self-

Dealing Index and the Rule of Law) the relation between and among firm characteristics,

compensation structure, and the legal environment is consistent with theory and with prior

empirical evidence in Bryan et al. (2010).16

Moreover, regarding the focus of our study, the initial findings for the relation between

cultural determinants and compensation structure are very supportive of theory (and our

hypotheses). Firms from countries with high levels of Individualism (Quartiles 3 and 4) use

more equity in their compensation structure. The data also support the theory regarding the

relation between UAI and equity-based compensation. That is, as expected, firms from countries

with low levels of UAI use larger relative amounts of equity-based compensation.

IV. Empirical Results - Determinants of CEO Compensation Structure

The following sections describe our empirical findings regarding the determinants of the

observed cross-sectional differences in the use of equity-based compensation. Our regressions

include industry and year dummy variables to control for fixed effects and firm-specific

variables that proxy for agency costs. We use a Tobit model as the data are left censored. The

dependent variable is the percent of equity in the compensation structure (EquityPct); the

independent variables of primary interest are measures of the cultural environment. As is

general practice with panel data, we correct the standard errors for heteroscedasticity and serial

correlation.

A. The Legal Environment and Compensation Structure

Before testing whether culture affects compensation structure, we first confirm the

16 Bryan et al. (2010) examine similar data but only for 1996-2000.

25

empirical relation between legal environment and the use of equity-based compensation.

Accordingly, our models include measures of the issuer’s home legal system.

Table IV about here

The results in Table IV are consistent with our expectations and with earlier findings by

Bryan et. al. (2010) for data over the 1996-2000 period, and by Bryan et al. (2011) for the 1996-

2008 period. After controlling for industry and year fixed effects, the Anti Self-Dealing Index is

significantly positively related to the proportion of equity in the compensation structure. This is

true for the sample that includes U.S. firms (Model 1) and for the sample of non-U.S. firms

(Model 3). Similarly, the Rule of Law is also significantly positively related to the proportion of

equity in the compensation structure for all firms and for the subsample of non-U.S. firms.17

B. The Cultural Environment and Compensation Structure

We examine the impact of culture on compensation structure in Table V. In Panel A, we

investigate the full sample (that includes U.S. firms). In Panel B, we restrict our analysis to non-

U.S. firms. In both panels, Models 1 to 3 (4 to 6) examine the relation between a country’s level

of Individualism (UAI) and the structure of the compensation package. Models 2 and 5 also

control for industry and year fixed effects, while Models 3 and 6 add firm-level characteristics

(size, the market-to-book ratio, leverage, and free cash flow) as further control variables.

The results in Panels A (all firms) and B (non-U.S. firms) of Table V indicate that,

consistent with our hypothesis, culture plays an important role in explaining cross-sectional

17 The empirical analysis in Table IV also confirms the significance of firm-specific determinants of compensation structure. With respect to the agency factors reflected by enterprise-level characteristics, we find that firms that are larger, have higher market-to-book ratios, and are less levered are significantly more likely to use equity-based compensation. For non-U.S. firms, in addition to the significant relations mentioned above, free cash flow is also negatively related to the use of equity in the compensation structure. These results are all consistent with prior studies of the compensation structures of U.S. firms (Yermack, 1995) and with the evidence in Bryan et al. (2010, 2011) for U.S. and non-U.S. firms.

26

differences in compensation structures. In Models 1 to 3 of both panels, Individualism is

significantly positively related (at the one percent level) to the use of equity-based compensation

for both the full sample and the subsample of only non-U.S. firms. The results also indicate that

Uncertainty Avoidance is significantly negatively related (at the one percent level) to the equity

component of the compensation for both the full sample and the subsample of only non-U.S.

firms. The magnitude of the coefficients on Individualism and Uncertainty Avoidance are

similarly significant for the sample of all firms and for the subsample of non-U.S. firms.

Overall, the findings in Table V provide strong support for the hypothesis that the

cultural environment, as captured by Individualism and Uncertainty Avoidance, is significantly

related to the cross-sectional variation in compensation structures. These results for our cultural

variables are highly significant after controlling for firm-level characteristics and for industry

and year fixed effects.18

C. Cultural and Legal Environments as Determinants of Compensation Structure

Having determined that proxies for the legal environment and cultural environment

separately explain cross-sectional variation in compensation structure, we next examine the

correlation matrix for the cultural and legal variables. We then provide the results for our models

which include both sets of factors.

Table VI presents the correlation matrix for the two legal variables (Anti Self-Dealing

Index and the Rule of Law) and our primary cultural variables (Individualism and Uncertainty

Avoidance). Individualism is significantly positively correlated with both the Anti Self-Dealing

Index and the Rule of Law. This is consistent with the notion that highly individualistic societies

have stronger protection of minority shareholders and more effective enforcement of laws. 18 In models containing each cultural variable, we find that the enterprise-specific control variables (measuring likelihood of agency conflicts) remain significantly related to compensation structure.

27

Uncertainty Avoidance is negatively correlated with both variables representing the country’s

legal environment.

Insert Table VI about here

Since we focus on how culture affects the cross-sectional differences in compensation

structure, our robustness tests in the next section present several alternative variables that the

literature identifies as capturing different aspects of a nation’s cultural environment. We also

include those variables in this correlation matrix. Specifically, Table VI includes Schwartz’s

measure of Harmony (whose polar opposite is Mastery), the percentage of respondents in a

country who are Catholic (Catholic), the extent to which respondents of the World Values

Survey report a distrust of companies in their country (DistrustComp), and a dummy variable

that takes on a value of 1 if the primary language of the country is English (LangEng).19

Table VII presents findings for the multivariate regressions that consider both the cultural

and legal variables. Models 1 to 4 investigate the sample that includes U.S. firms, while Models

5 to 8 focus on the non-U.S. sample. Models 1 and 5 examine whether the two legal variables

remain significant after controlling for industry and year fixed effects and firm-specific

characteristics, but without either of the cultural variables. Models 2 and 6 add Individualism as

the cultural variable, while Models 3 and 7 add Uncertainty Avoidance as the cultural variable.

Models 4 and 8 of Table VII include both Individualism and Uncertainty Avoidance in the

regressions.

Table VII about here

19 In Section I, we described each of these complementary cultural variables and the expected relation of each to compensation structure.

28

In considering the impact of culture on compensation design, the primary variables of

interest are Hofstede’s measures of Individualism and Uncertainty Avoidance. The data strongly

indicate that culture is a significant determinant of compensation structure. After controlling for

the legal environment and for firm-specific agency variables (and industry and year fixed

effects), we confirm that culture matters. Individualism, in Models 2 and 6, is significantly

positively related to equity in the compensation mix. Similarly, the coefficient on Uncertainty

Avoidance, in Models 3 and 7, is negative and significant at the one percent level. Furthermore,

both cultural variables remain significant in Model 4. This indicates that the cultural dimensions

reflected by Individualism and Uncertainty Avoidance are each significant determinants of

compensation structure, and that culture reflects aspects of a country’s institutional setting that

are different from those measured by the legal environment.20

Models 5 to 8 of Table VII verify that the cultural dimensions of a country are

significantly related to its pay practices. Individualism and Uncertainty Avoidance are significant

in a majority of the regression models. These tests confirm the importance of culture in

explaining the use of equity-based compensation (after accounting for the other, previously-

identified determinants of compensation structure).

V. Robustness Tests - Potential Omitted Variables Problem

We recognize that our findings of a relation between culture and compensation structure

could be due to an omitted variable that is correlated with both culture and compensation

20 The results also indicate that both the Anti Self-Dealing Index and the Rule of Law are significantly positively related to the proportion of equity in the compensation mix. Both variables are significant at the one percent level in Models 1 through 4. In Models 5 through 8, the results are slightly weaker for the Rule of Law, but continue to indicate that the proportion of equity-based compensation is significantly related to the legal environment in the non-U.S. sample. The results further show that the firm-specific variables have significant explanatory power. This finding is consistent with theory and with previous findings in Bryan et al. (2010, 2011).

29

structure. We perform the following additional tests to mitigate concerns about this potential

problem.21

A. Simulated Slope Coefficients

To investigate whether our results are subject to an omitted correlated variables problem,

we re-run the regressions in Models 4 and 8 of Table VII after randomly assigning values for

Individualism and Uncertainty Avoidance to each country. We do not change the data for the

remaining legal and control variables. We estimate a distribution of slope coefficients for the two

cultural factors by performing the regressions 1000 times. We then test whether the slope

coefficients reported in Models 4 and 8 of Table VII are significantly different from zero.

Focusing on the sample that includes U.S. firms, we find that the mean (median) value of

the slope coefficient on Individualism is 0.00014 (0.00054) with a standard deviation of 0.0050.

This indicates that we are unable to reject the hypothesis that the average (median) slope

coefficient is different from zero when the data are randomly generated. However, the slope

coefficient of 0.0123 in Model 4 of Table VII is significantly positive at the one percent level.

For Uncertainty Avoidance, the mean (median) slope coefficient is 0.00018 (0.00037) with a

standard deviation of 0.00434. This indicates that the average (median) coefficient is not

different from zero. The value of -0.0017 in Model 4 of Table VII is also insignificant. Together,

the results from the simulated distribution of slope coefficients indicate that culture, most

specifically Individualism, is significantly related to compensation structure.

When we consider only the non-U.S. firms, the simulated mean (median) value of the

slope coefficient on Individualism is -0.00008 (-0.00003) with a standard deviation of 0.00425.

This indicates that the mean (median) simulated value is not significantly different from zero.

21 The authors thank Mike Lemmon for these suggestions.

30

However, the reported value of 0.0133 in Model 8 of Table VII is significantly positive at the

one percent level. Regarding Uncertainty Avoidance, the mean (median) value is 0.00020

(0.00026) with a standard deviation of 0.00354, indicating that the value is not different from

zero. The value reported in Table VII, -0.0017, is also insignificant. Overall, this finding that the

“randomized” Hofstede values are not significant in the simulated models provides some

assurance that the cultural variables in the primary models are not simply capturing a country

effect.

B. Tradable versus Non-Tradable Goods

If culture does contribute to the variation in compensation structure across countries, its

importance should depend on whether the labor market for CEOs is global versus local. If CEOs

are hired locally, compensation structures should be more dependent on local norms. However, if

CEOs are hired from the global labor market, local norms (i.e., culture) should play less of a role

in explaining the variation in compensation structures. We hypothesize that firms whose

products have a primarily domestic market are more likely to hire their CEOs from a local talent

pool. As such, we expect national culture to play a more significant role in explaining the cross-

sectional variation in compensation structure for firms with non-tradable (or domestic) goods

versus those whose products are traded internationally.

We base this hypothesis on the “organizational learning” literature. For example,

Burkema and Vermaleun (1998) and Kogut and Zander (1993) argue that firms operating in

diverse, multi-national markets (such as firms that we distinguish as selling “tradable” goods)

develop a more cosmopolitan orientation brought on by broader exposures of ways of doing

things in other cultures. If a firm functions in a purely domestic market (such as the firms that

we designate as selling “non-tradable” goods), its more homogenous environment limits

31

opportunities to learn from the outside. As a result, the policies and processes of the more

isolated firm will be more likely to primarily reflect its local culture.

We create an indicator variable, Nontrade, that takes on a value of 1 if the firm produces

goods that are likely to be traded only in the firm’s domestic market, and 0 if the firm produces

goods that are likely to trade internationally. The variable specification follows Sarkissian and

Schill (2004).22 We feel that using tradable versus non-tradable goods is a reasonable means of

distinguishing firms with a more domestic focus from those with a more multinational

orientation. Dreher (2006), Clark (2000), and Norris (2000) note that the cross-regional flow of

goods helps to erode national boundaries and mediates cultural differences. Therefore, the

decision-making and contracting policies for a firm producing tradable goods should be less

impacted by its local culture. Accordingly, for firms that are more cosmopolitan, the measures

of national culture should have a less pronounced effect on compensation structure.

Our Tobit model includes interaction terms for all independent variables to allow the

slope coefficients to differ across the two subsamples. The primary variables of interest are the

coefficients on the two culture variables and the coefficients on the interaction terms between the

two culture variables and the indicator variable. The slope coefficients on the culture variables

capture the relation between culture and compensation structure for firms whose goods are

traded internationally (tradable goods). The coefficients on the interaction terms capture whether

the relation between culture and compensation is significantly different for those firms whose

22 Sarkissian and Schill (2004) use industry classifications to distinguish firms producing tradable and non-tradable goods. Industries with tradable goods are: chemicals, consumer products, electronics, manufacturing, pharmaceuticals, mining, oil and gas, and paper. Coval and Moskowitz (2001) and Kang and Stulz (1997) similarly partition firms by output type (tradable vs. non-tradable) when examining the geographical preferences of equity investors.

32

goods trade in domestic markets (non-tradable goods). Our hypothesis is that the slope

coefficient on the interaction terms will be significantly different from zero.23

Insert Table VIII about here

Models 1-3 of Table VIII provide results for the full sample of firms, while Models 4-6

provide results for the sample of non-U.S. firms. The coefficient on Individualism is significantly

positive at the one percent level in all of the models. The coefficient on Uncertainty Avoidance

is significant for the full sample of firms. Together, the results indicate that culture (most

specifically, Individualism) does explain the variation in compensation structure for firms whose

goods are traded internationally.

Consistent with our hypothesis, the coefficient on the interaction term with Individualism

is significantly positive at the one percent level in all models. The coefficient on the interaction

term with Uncertainty Avoidance, however, is insignificant at conventional levels. Overall, the

findings are consistent with the hypothesis that culture is more significant in explaining the

variation in compensation structure for firms whose goods trade primarily in domestic markets.

Since we expect firms whose products trade in domestic markets may be more likely to hire

CEOs primarily from their local markets, our findings suggest that local norms may exert a

stronger influence over compensation structure for these firms. Furthermore, our results indicate

that culture does explain, albeit less strongly, the variation in compensation structure for firms

with global products.

VI. The Relative Importance of Culture and Law

23 This approach is also used by Stulz and Williamson (2003) and Rameriz and Tadesse (2007). Using similar types of interaction terms, both studies conclude that a greater international orientation will temper the impact of national culture while the impact of culture is more pronounced when firms are more domestically focused.

33

The results thus far indicate that both culture and the legal environment are related to the

cross-sectional variation in the structure of CEO compensation. In this section, we investigate the

role of culture and legal environment in determining the unobservable time-invariant country

level heterogeneity (or country level fixed effect) in the cross-sectional variation in

compensation structure.24 Specifically, we are interested in understanding what percentage of the

country fixed effects is explained by the cultural and legal variables that we study.

We begin by estimating a Tobit model where the dependent variable is the relative use of

equity based pay (EquityPct) and the independent variables are the enterprise-specific factors

examined in Tables V and VII (firm size, market-to-book ratio, leverage, and free cash flow). In

this model, we include firm fixed effects to capture any unobservable time-invariant firm-level

heterogeneities. We also include country fixed effects in the model. The coefficient estimates on

the country fixed effects then become the variable of interest in the second stage of the analysis.

Here, we investigate whether Individualism, Uncertainty Avoidance, the Anti Self-Dealing Index,

and the Rule of Law are related to the country fixed effects.

Table IX presents findings for the regressions explaining the unobservable time-invariant

country level heterogeneities. The coefficients on all variables from the univariate OLS

regressions have the expected signs. Model 1 indicates that Individualism is significantly

positively related to, and explains 25% of, the country fixed effects. Uncertainty Avoidance, is

significantly negatively related to, but explains only 13% of the country fixed effects. Both the

Anti-Self Dealing Index and the Rule of Law are significantly positively related to, but explain

less than 8.5% of, the variation in country fixed effects.

24 Graham, Li, and Qui (2011) use this methodology to examine the role of unobservable time-invariant firm and manager heterogeneities in determining executive pay.

34

Table IX about here

The multivariate regressions in Models 5 through 7 of Table IX continue to provide

strong support that Individualism is important in explaining the unobservable time-invariant

country level heterogeneity in compensation structure. The Individualism variable is significantly

positive in all three multivariate regressions. The contribution of Uncertainty Avoidance is

smaller (though still marginally significant in Model 5). The variables representing the legal

environment provide no marginal contribution to what is already captured by the two cultural

variables. While Individualism alone explains 25% of the country fixed effects, the two cultural

variables combined explain about 30% of the variation in county fixed effects, suggesting that

Uncertainty Avoidance does add some explanatory power above that provided by Individualism.

However, adding the legal variables provides no additional explanatory power.

In summary, these results indicate that culture appears to be more important than law in

explaining the unobservable time-invariant country level heterogeneity in compensation

structure. Second, for the firms in our sample, Individualism alone captures most of the

explanatory power provided by the cultural environment. Third, additional work is needed to

understand the unexplained portion of the unobservable time-invariant country level

heterogeneity in compensation structure.25 Overall, the findings from Table IX strongly suggest

that understanding culture may substantially contribute to understanding compensation structure.

VII. Robustness Tests - Alternative Instruments for Culture

25 For example, taxes may play a role in the cross-sectional variation of executive compensation. Accordingly, we also estimated the models in Table IX after controlling for the impact of taxes. While there are many tax issues to consider across 43 countries and throughout 14 years (1996-2009), we feel that the tax differential most relevant to compensation structure would be the country-specific spread between taxes on capital gains and ordinary income. We refer to this spread as a “tax wedge”. We gathered data for capital gains tax rates for each country from Djankov et al. (2010). We gathered data for the highest marginal tax rate on ordinary income in each country from the Financial Structure and Economic Development Database from the World Bank (Beck et al. (2000)). Our measure of the “tax wedge” was insignificant in all models.

35

In this section, we draw from the extant literature to identify alternative instruments of

culture (but also include proxies to control for the legal environment, for firm-specific agency

costs, and for industry and year fixed effects). Panel A of Table X provides our findings for the

sample that includes U.S. firms, while Panel B presents our results for non-U.S. firms.

A. Schwartz’s Harmony Metric and the Structure of Compensation

As discussed previously, Schwartz’s (2006) measures are seen as alternatives to

Hofstede’s statistics in the cross-cultural psychology literature. We investigate Schwartz’s

measure of Harmony (whose polar opposite is Mastery). The Schwartz Mastery/Harmony metric

captures certain cultural elements (which may also affect the stockholder/manager agency

conflicts) that are less explicitly measured by Hofstede’s Individualism statistic. We hypothesize

that Harmony will be negatively related to the use of equity-based compensation.

Model 1 in Panels A and B of Table X indicate that, consistent with our expectations,

Harmony is negatively related to the proportion of equity in the compensation mix, even after

controlling for the legal environment. The coefficient is significant at the one percent level, as

are the coefficients of the two legal variables.

Table X about here

B. Religion and the Structure of Compensation

Guiso et al. (2003, 2006, and 2008) and Stulz and Williamson (2003), among others,

have also used religion to measure aspects of a country’s culture. Accordingly, we use the

proportion of respondents to the World Values Survey, in each country, who stated that they

were Catholic. As noted earlier, we expect that countries with higher proportions of Catholics

should use less equity-based compensation.

36

Model 2 (Panels A and B) of Table X indicates that, consistent with our hypothesis, firms

from countries with higher percentages of Catholics use less equity in the compensation

structure. The coefficient is significant at the one percent level for the U.S. sample (Panel A),

and significant at the five percent level when studying non-U.S. firms (Panel B). The regression

coefficients for legal environment (Anti Self-Dealing Index and the Rule of Law) are positive and

significant at the one percent level in both Panels A and B. This suggests that religion is another

cultural factor that affects compensation structure (even after controlling for law).

C. Language and the Structure of Compensation

Following Stulz and Williamson (2003) and Licht et al. (2001), we consider language as

a measure of culture. Specifically, we investigate whether countries where the dominant

language is English have firms that use more equity-based compensation (after controlling for

the legal environment, for firm-specific agency factors, and for industry and year fixed effects).

Model 3 (Panels A and B) of Table X present our findings. The data indicate that when

examining the U.S. sample (Panel A), the proportion of equity-based compensation is

significantly higher for firms in countries where English is the dominant language. Our results

are weaker when examining non-U.S. firms. Also, in both Panels A and B, our measures of the

country’s legal environment remain significant (indicating that language adds an independent

dimension when explaining the cross-sectional variation in compensation structure).

D. Trust and the Structure of Compensation

We hypothesize that trust should be related to pay practices across countries. Guiso et al.

(2008) find that individuals hold less wealth in the stock market in countries with greater degrees

of distrust (which contributes to lower levels of equity market development). Bryan et al. (2010)

document that firms use less equity in the compensation mix in countries where the stock market

37

is less developed. As such, we hypothesize a negative relation between trust and the use of

equity-based compensation. We follow Aghion et al. (2010) and focus on the level of distrust of

companies in a country. Since higher levels of distrust should yield less equity investment by

minority shareholders and a less developed equity market, we expect a lower use of equity-based

compensation in countries with higher levels of distrust of companies.

Panels A and B of Table X indicate that, consistent with our expectations, firms in

countries with higher levels of distrust use less equity-based compensation. Our findings are

significant at the one percent level when the sample includes U.S. firms, and significant at the

ten percent level when examining only non-U.S. firms.

Overall, we find that alternative specifications of national culture are also significantly

related to the cross-sectional variation in the compensation structures of firms. Since these

findings exist after controlling for previously-identified determinants of compensation structure

(i.e., firm-specific factors and legal environment), the data strongly suggest that culture plays a

separate and significant role in the design of executive compensation contracts.

VIII. Summary and Conclusions

Are the forces of globalization mitigating the effects of cross-national differences so that

the world is becoming “flat” and that the most effective CEO compensation structures are

becoming more widely-adopted? In other words, are cross-country differences in CEO

compensation structures dissipating over time?

Our study addresses this issue by using CEO compensation data for firms from 43

countries over the 1996 to 2009 period. Instead of finding that cross-country differences in

compensation structures are dissipating over time, we document that CEO compensation

structures across countries, specifically the relative use of equity-based pay, have not converged.

38

While the average structure of CEO compensation differs for firms segregated by legal origin,

the time series trend in these data does not converge over time. This indicates that country-level

factors other than legal environment must affect the cross-sectional differences in CEO

compensation structure.

We add to the literature by examining whether culture contributes to persistent

differences in compensation structures across countries. We rely on dimensions of culture

captured by Hofstede (2001), Schwartz (2006), and the World Values Survey. Our study

expands the application of quantitative measures of cultural dimensions to the financial

contracting literature and provides initial evidence of how national culture impacts CEO

compensation structure.

We find that culture matters. Our results suggest that cross-national differences in

compensation structures are significantly related to cross-national differences in culture.

Specifically, in this study, we focus on the cultural value dimensions that should most directly

affect agency conflicts and thus should most directly affect the structure of executive

compensation. The data indicate that these measures of culture are significant determinants of

the use of equity-based compensation. Consistent with our expectations, we show that the equity

portion of CEO compensation is positively related to a nation’s level of Individualism and

negatively related to its degree of Uncertainty Avoidance. These significant results remain after

controlling for legal environment, firm-specific agency factors, and industry and year fixed

effects. We also present evidence to mitigate concerns regarding problems potentially stemming

from endogeneity and omitted variables. Moreover, our robustness tests help to confirm that

alternative instruments for national culture (e.g., language, religion, trust) play a significant role

in explaining cross-national differences in compensation structures. Overall, by verifying the

39

strong relation between culture and CEO compensation design, we contribute to a more

comprehensive understanding of the interaction between informal institutions (i.e., national

culture) and corporate decision-making.

40

REFERENCES Aghion, P., Y. Algan, P. Cahuc, and A. Shleifer, 2010, “Regulation and Distrust”, Quarterly Journal of Economics 125, 1015-1049. Ahern, K., D. Daminelli, and C. Fracassi, 2012, “Lost in Translation? The Effect of Cultural Values on Mergers Around the World”, Journal of Financial Economics, forthcoming. Baker, G., and B. Hall, 2004, “CEO Incentives and Firm Size”, Journal of Labor Economics 22, 767-798. Barkema, H., and F. Vermeulen, 1998, “International Expansion Through Start-Up or Acquisition: A Learning Perspective”, Academy of Management Journal 41, 7-26. Beck, T., A.Demirgüç-Kunt, and R. Levine, 2000, “A New Database on Financial Development and Structure”, World Bank Economic Review 14, 597-605. Bizjak, J., J. Brickley, and J. Coles, 1993, “Stock-Based Incentive Compensation and Investment Behavior”, Journal of Accounting and Economics 16, 349-372. Black, F., and M. Scholes, 1973, “The Pricing of Options and Corporate Liabilities”, Journal of Political Economy 81, 637-654. Booth, L., V. Aivazian, A. Demirguc-Kunt, and V. Maksimovic, 2001, “Capital Structure in Developing Countries”, Journal of Finance 56, 87-130. Breuer, W., and A. Salzman, 2010, “National Culture and Corporate Governance”, Working Paper, RWTH Aachen University. Bryan, S., L. Hwang, and S. Lilien, 2000, “CEO Stock-Based Compensation: An Empirical Analysis of Incentive-Intensity, Relative Mix, and Economic Determinants”, Journal of Business 73, 661-693. Bryan, S., R. Nash, and A. Patel, 2011, “Law and Executive Compensation”, Journal of Applied Corporate Finance 23, 84-91. Bryan, S., R. Nash, and A. Patel, 2010, “How Legal System Affects the Equity Mix in Executive Compensation”, Financial Management 39, 393-418. Bryan, S., R. Nash, and A. Patel, 2006, “Can the Agency Costs of Debt and Equity Explain the Changes in Executive Compensation During the 1990s?”, Journal of Corporate Finance 12, 516-535. Chui, A., A. Lloyd, and C. Kwok, 2002, “The Determination of Capital Structure: Is National Culture a Missing Piece to the Puzzle?”, Journal of International Business Studies 33, 99-127.

41

Clark, W., 2000, “Environmental Globalization”, in J. Nye and J. Donahue (eds.), Governance in a Globalizing World, Washington: Brookings Institution Press. Coase, R., 1937, “The Nature of the Firm”, Economica 4, 386-405. Coffee, J., 2001, “Do Norms Matter? A Cross-Country Evaluation”, University of Pennsylvania Law Review 149, 2151-2177. Coffee, J., 1999, “The Future as History: The Prospects for Global Convergence in Corporate Governance and Its Implications”, Northwestern University Law Review 93, 641-707. Coval, J., and T. Moskowitz, 2001, “The Geography of Investment: Informed Trading and Asset Prices”, Journal of Political Economy 109, 811-841. Desai, M., F. Foley, and J. Hines, 2004, “A Multinational Perspective on Capital Structure Choice and Internal Capital Markets”, Journal of Finance 59, 2451-2487. Djankov, S., T. Ganser, C. McLiesh, R. Ramalho, and A. Shleifer, 2010, “The Effect of Corporate Taxes on Investment and Entrepreneurship”, American Economic Journal: Macroeconomics 2, 31-64. Djankov, S., R. LaPorta, F. Lopez-de-Silanes, and A. Shleifer, 2008, “The Law and Economics of Self-Dealing”, Journal of Financial Economics 88, 430-465. Dyck, A., and L. Zingales, 2004, “Private Benefits of Control: An International Comparison”, Journal of Finance 59, 537-600. Fernandes, N., M. Ferreira, P. Matos, and K. Murphy, 2012, “Are U.S. CEOs Paid More? New International Evidence”, Review of Financial Studies, forthcoming. Friedman, T., 2005, The World is Flat: A Brief History of the Twenty-First Century, London: Straus and Giroux. Fukuyama, F., 1995, Trust, New York: Free Press. Gabaix, X., and A. Landier, 2008, “Why Has CEO Pay Increased So Much?”, Quarterly Journal of Economics 123, 49-100. Gaver, J., and K. Gaver, 1993, “Additional Evidence on the Association Between the Investment Opportunity Set and Corporate Financing, Dividend, and Compensation Policies”, Journal of Accounting and Economics 16, 125-160. Giannetti, M., and Y. Yafeh, 2012, “Do Cultural Differences Between Contracting Parties Matter? Evidence from Syndicated Bank Loans”, Management Science 58, 365-383.

42

Graham, J., S. Li, and J. Qiu, 2011, “Managerial Attributes and Executive Compensation”, Review of Financial Studies 25, 144-186. Griffin, D., K. Li, H. Yue, and L. Zhao, 2008, “Cultural Values and Corporate Risk-Taking”, Working Paper, University of British Columbia. Guiso, L., P. Sapienza, and L. Zingales, 2009, “Cultural Biases in Economic Exchange?”, Quarterly Journal of Economics 124, 1095-1131. Guiso, L., P. Sapienza, and L. Zingales, 2008, “Trusting the Stock Market”, Journal of Finance 63, 2557-2600. Guiso, L., P. Sapienza, and L. Zingales, 2006, “Does Culture Affect Economic Outcomes?”, Journal of Economic Perspectives 20, 23-48. Guiso, L., P. Sapienza, and L. Zingales, 2003, “People’s Opium? Religion and Economic Attitudes”, Journal of Monetary Economics 50, 225-282. Hallpike, C., 1986, The Principles of Social Evolution, Oxford: Clarendon Press. Hofstede, G., 2001, Culture’s Consequences: Comparing Values, Behaviors, Institutions and Organizations Across Nations, Thousand Oaks: Sage Publications. Hofstede, G., 1991, Cultures and Organizations: Software of the Mind, London: McGraw-Hill. Hofstede, G., 1983, “The Cultural Relativity of Organizational Practices and Theories”, Journal of International Business Studies 14, 75-89. Hofstede, G., 1980, Culture’s Consequences: International Differences in Work-Related Values, Thousand Oaks: Sage Publications. Iannaccone, L., 1998, “Introduction to the Economics of Religion”, Journal of Economic Literature 36, 1465-1495. Jensen, M., 1986, “Agency Costs of Free Cash Flow, Corporate Finance and Takeovers”, American Economic Review 76, 323-339. Jensen, M., and W. Meckling, 1976, “Theory of the Firm: Managerial Behavior, Agency Costs and Ownership Structure”, Journal of Financial Economics 3, 305-360. John, T., and K. John, 1993, “Top Management Compensation and Capital Structure”, Journal of Finance 48, 949-974. Johnson, S., R. LaPorta, F. Lopez-de-Silanes, and A. Shleifer, 2000, “Tunneling”, American Economic Review 90, 22-27.

43

Kang, J., and R. Stulz, 1997, “Why Is There a Home Bias? An Analysis of Foreign Portfolio Equity Ownership in Japan”, Journal of Financial Economics 46, 3-28. Kirkman, B., K. Lowe, and C. Gibson, 2006, “A Quarter Century of Culture’s Consequences: A Review of Empirical Research Incorporating Hofstede’s Cultural Values Framework”, Journal of International Business Studies 37, 285-320. Kogut, B., and U. Zander, 1993, “Knowledge of the Firm and the Evolutionary Theory of the Multinational Corporation”, Journal of International Business Studies 24, 625-645. Kogut, B., and H. Singh, 1988, “The Effect of National Culture on the Choice of Entry Mode”, Journal of International Business Studies 19, 411-432. Kwok, C., and S. Tadesse, 2006, “National Culture and Financial Systems”, Journal of International Business Studies 37, 227-247. La Porta, R., F. López-de-Silanes, A. Shleifer, and R.Vishny, 1997a, “Legal Determinants of External Finance”, Journal of Finance 52, 1131-1150. La Porta, R., F. López-de-Silanes, A. Shleifer, and R. Vishny, 1997b, “Trust in Large Organizations”, AEA Papers and Proceedings 87, 333-338. Lang, M., K. Lins, and D. Miller, 2003, “ADRs, Analysts, and Accuracy: Does Cross-Listing in the United States Improve a Firm’s Information Environment and Increase Market Value?”, Journal of Accounting Research 41, 317-345. Lehn, K., and A. Poulsen, 1989, “Free Cash Flow and Stockholder Gains in Going Private Transactions”, Journal of Finance 44, 771-787. Licht, A., C. Goldschmidt, and S. Schwartz, 2001, “Culture, Law, and Finance: Cultural Dimensions of Corporate Governance Laws”, Working paper, Interdisciplinary Center (Herzliya, Israel). Megginson, W., R. Nash, and M. Van Randenborgh, 1994, “The Financial and Operating Performance of Newly-Privatized Firms: An International Empirical Analysis”, Journal of Finance 49, 403-452. Norris, P., 2000, “Global Governance and Cosmopolitan Citizens”, in J. Nye and J. Donahue (eds.), Governance in a Globalizing World, Washington: Brookings Institution Press. North, D., 1990, Institutions, Institutional Change, and Economic Performance, Cambridge: Cambridge University Press. Pagano, M., A. Roell, and J. Zechner, 2002, “The Geography of Equity Listing: Why Do Countries List Abroad?”, Journal of Finance 57, 2651-2694.

44

Rajan, R., and L. Zingales, 1995, “What Do We Know About Capital Structure? Some Evidence from International Data”, Journal of Finance 50, 1421-1460. Ramirez, A., and S. Tadesse, 2007, “Corporate Cash Holdings, National Culture, and Multinationality”, Working paper, University of Michigan. Rauch, J., 1999, “Networks Versus Markets in International Trade”, Journal of International Economics 41, 7-35. Reese, W., and M. Weisbach, 2002, “Protection of Minority Shareholder Interests, Cross-Listings in the United States, and Subsequent Equity Offerings”, Journal of Financial Economics 66, 65-104. Sapienza, P., A. Toldra, and L. Zingales, 2010, “Understanding Trust”, Working Paper, Northwestern University. Sarkissian, S, and M. Schill, 2004, “The Overseas Listing Decision: New Evidence of Proximity Preference”, Review of Financial Studies 17, 769-810. Schwartz, S., 2006, “Value Orientations: Measurement, Antecedents and Consequences Across Nations”, in R. Jowell, C. Roberts, R. Fitzgerald, and G. Eva (eds.), Measuring Attitudes Cross-Nationally - Lessons from the European Social Survey, London: Sage Publications. Siegel, J., A. Licht, and S. Schwartz, 2011, “Egalitarianism and International Investment”, Journal of Financial Economics 102, 621-642. Sondergaard, M., 1994, “Research Note: Hofstede’s Consequences: A Study of Reviews, Citations, and Replications”, Organization Studies 15, 447-456. Stulz, R., and R. Williamson, 2003, “Culture, Openness and Finance”, Journal of Financial Economics 70, 313-349. Tesar, L., and I. Werner, 1995, “Home Bias and High Turnover”, Journal of International Money and Finance 15, 467-492. Yermack, D., 1995, “Do Corporations Award CEO Stock Options Effectively?”, Journal of Financial Economics 39, 237-269.

45

Table I Observations for Each Country by Year

This table presents the number of firm-level observations for which we have executive compensation data. The data are presented by country for the years 1996-2009. We identify the American Depository Receipts (ADR) issuers from Compustat and obtain the specific executive compensation data from Form 20-F, provided by the SEC’s EDGAR database. We express the observations in firm years.

Year Country 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 TotalsArgentina 4 7 8 8 9 8 9 9 9 5 4 4 3 7 94Australia 2 7 9 10 7 11 10 10 9 6 5 6 5 3 100Austria 1 1Belgium 1 1 1 1 1 1 1 1 1 1 10Bermuda 1 2 4 4 2 1 29 43Brazil 1 3 10 16 16 16 16 16 17 8 7 7 6 25 164British Virgin Islands 1 1 1 1 4 8Cayman Islands 1 23 27 51Chile 8 10 11 11 12 12 12 12 11 7 6 6 5 8 131China 5 7 7 8 10 8 9 9 9 8 7 7 7 15 116Denmark 1 1 1 1 1 1 1 1 1 1 1 1 1 13England 14 23 33 40 53 52 50 50 51 29 27 27 38 19 506Finland 1 1 1 3 6 1 1 1 1 1 1 1 1 20France 9 14 18 17 19 15 14 14 13 13 12 12 14 6 190Germany 5 5 5 6 11 9 8 7 6 4 4 4 5 2 81Greece 1 2 2 1 1 1 1 3 12Hong Kong 1 1 2 4 6 1 1 1 1 1 1 1 1 3 25Hungary 1 1 1 1 4India 1 2 5 3 3 3 4 3 3 3 3 2 35Indonesia 1 2 2 1 1 1 1 2 2 2 1 16Ireland 1 5 8 9 12 9 9 9 8 4 3 3 3 7 90Israel 5 7 7 7 10 7 7 6 7 5 5 5 5 2 85Italy 4 5 8 10 9 10 10 9 11 8 7 8 9 2 110

46

Table I (continued)

Year Country 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 TotalsJapan 4 9 10 13 14 10 9 9 11 8 7 8 7 16 135Luxembourg 1 2 2 2 3 3 13Mexico 11 13 16 21 17 11 10 11 13 8 8 8 7 12 166Netherlands 9 11 16 16 17 14 14 14 14 7 6 6 5 10 159New Zealand 1 2 2 2 1 1 1 1 1 1 1 1 1 16Norway 2 2 1 1 2 1 1 1 1 12Peru 1 1 1 1 1 1 1 7Philippines 1 1 1 2 1 1 1 1 1 1 1 1 1 14Portugal 1 1 2 1 1 1 1 1 1 1 1 1 13Russia 1 1 2 3 2 2 2 2 2 2 2 2 2 4 29Singapore 1 2 1 1 1 1 1 1 1 1 11South Africa 2 3 3 6 4 3 3 3 3 3 3 3 5 44South Korea 3 3 3 5 3 4 4 4 4 3 3 3 3 8 53Spain 1 2 3 4 5 3 3 3 3 1 1 1 1 3 34Sweden 5 8 8 9 9 8 7 6 6 4 3 4 5 82Switzerland 1 1 2 1 7 5 5 5 5 3 2 2 2 9 50Taiwan 1 1 1 4 3 2 3 3 3 2 2 2 5 32Turkey 1 1 1 1 1 1 6Venezuela 2 2 2 2 1 1 1 2 2 2 1 2 20Non U.S. Totals

102 160 210 249 292 239 230 228 235 155 139 141 173 248 2,801

U.S. Firms 1,661 1,680 1,736 1,809 1,785 1,678 1,678 1,750 1,754 1,750 1,857 1,926 1,833 1,747 24,644

All Firms 1,763 1,840 1,946 2,058 2,077 1,917 1,908 1,978 1,989 1,905 1,996 2,067 2,006 1,995 27,445

47

Table II

Mean Values of Compensation Structure and Firm-Specific Variables by Country

This table presents the mean values for our measure of compensation structure (EquityPct) and for our firm-specific control variables. EquityPct is the ratio of total equity-based compensation (stock option compensation plus restricted stock compensation) to total compensation. Total compensation is the sum of cash compensation, stock option compensation, restricted stock, and long-term incentive compensation. We measure stock option compensation with the Black-Scholes (1973) option pricing model. Size is the natural logarithm of the firm’s total assets. Market-to-Book (MVBV) is the book value of total assets less the book value of equity plus the market value of equity divided by the book value of total assets. Leverage (LEV) is the ratio of the firm’s total debt to total assets. Free Cash Flow (FCF) is the ratio of operating income before depreciation less the sum of income tax, interest, and dividends paid scaled by the firm’s market value of equity and total debt. Mean

Country Obs EquityPct Size MVBV LEV FCFArgentina 94 0.000 7.071 1.847 0.288 0.071Australia 100 0.334 6.407 2.429 0.202 0.002Austria 1 0.000 7.860 1.408 0.566 0.016Belgium 10 0.185 9.399 1.109 0.422 0.094Bermuda 43 0.443 7.524 1.077 0.303 -0.168Brazil 164 0.039 8.481 1.067 0.361 0.123British Virgin Islands 8 0.245 6.032 1.567 0.090 -0.063Cayman Islands 51 0.109 6.097 1.512 0.180 0.055Chile 131 0.006 7.476 1.508 0.370 0.035China 116 0.052 8.767 1.255 0.285 0.075Denmark 13 0.000 8.682 3.252 0.120 0.044England 506 0.256 8.437 2.265 0.297 0.007Finland 20 0.244 9.034 2.839 0.122 0.037France 190 0.155 7.172 2.101 0.210 0.010Germany 81 0.163 3.746 1.935 0.118 0.032Greece 12 0.071 8.990 1.284 0.408 0.099Hong Kong 25 0.050 9.412 2.143 0.131 0.093Hungary 4 0.002 8.648 1.789 0.532 0.112India 35 0.103 6.318 2.817 0.109 0.015Indonesia 16 0.000 7.751 1.707 0.388 0.069Ireland 90 0.139 6.525 1.952 0.214 0.012

48

Table II (continued)

Mean

Country Obs EquityPct Size MVBV LEV FCFIsrael 85 0.153 6.413 1.867 0.266 0.019Italy 110 0.053 8.399 1.491 0.311 0.062Japan 135 0.038 9.919 1.321 0.283 0.069Luxembourg 13 0.497 10.576 1.266 0.312 0.081Mexico 166 0.012 7.810 1.418 0.329 0.068Netherlands 159 0.152 9.906 1.510 0.402 0.044New Zealand 16 0.350 8.342 2.143 0.472 0.057Norway 12 0.089 6.309 1.593 0.191 0.018Peru 7 0.000 7.960 1.646 0.367 0.117Philippines 14 0.000 8.630 1.653 0.470 0.063Portugal 13 0.000 9.759 1.419 0.495 0.069Russia 29 0.053 8.035 2.273 0.384 0.088Singapore 11 0.278 8.505 1.311 0.419 0.062South Africa 44 0.137 7.293 1.675 0.393 0.020South Korea 53 0.043 10.212 1.236 0.373 0.131Spain 34 0.021 10.847 1.506 0.450 0.085Sweden 82 0.028 6.499 2.365 0.254 0.010Switzerland 50 0.281 9.569 2.217 0.258 0.045Taiwan 32 0.091 8.946 2.342 0.220 0.087Turkey 6 0.000 8.544 2.300 0.258 0.065Venezuela 20 0.000 7.260 1.706 0.131 0.142U.S. 24,644 0.430 7.528 1.857 0.292 0.028

49

Table III Mean Values of Compensation Structure and Firm-Specific Variables

by Potential Legal and Cultural DeterminantsThis table presents the mean values of compensation structure (EquityPct) and the main control variables arrayed according to factors measuring the nation’s legal environment (SelfDeal and Rule of Law) and cultural environment (Individualism, IDV; and Uncertainty Avoidance, UAI). SelfDeal is a measure of shareholder rights protection (from Djankov et al. (2008)), where higher values suggest stronger protection of shareholder rights. Rule of Law (from La Porta et al. (1997a)) is a measure of how effectively a nation enforces its laws. Higher values indicate a stronger tradition of enforcement. IDV (from Hofstede (1980)) reflects the relation between the individual and the group. Higher scores indicate a culture in which individual interests are more likely to prevail over collective interests. UAI (from Hofstede (1980)) measures the degree to which people feel stress or discomfort in unstructured or risky circumstances. Higher values indicate less tolerance of risk. Quartiles are based on the number of countries in the sample. EquityPct is percentage of total compensation composed of stock options and restricted stock. Size is the natural logarithm of the firm’s total assets. Market-to-Book (MVBV) is the book value of total assets less the book value of equity plus the market value of equity divided by the book value of total assets. Leverage (LEV) is the ratio of the firm’s total debt to total assets. Free Cash Flow (FCF) is operating income before depreciation less the sum of income tax, interest, and dividends paid scaled by the firm’s market value of equity and total debt.

Panel A: Legal Environment: SelfDeal and Rule of Law SelfDeal

Countries Quartile EquityPct Size MVBV LEV FCF 11 1 7.31% 8.660 1.425 0.347 0.084 10 2 6.72% 7.482 1.868 0.271 0.039 11 3 4.80% 8.843 1.683 0.303 0.063 11 4 42.36% 7.534 1.859 0.292 0.028

Rule of Law Countries Quartile EquityPct Size MVBV LEV FCF

11 1 8.54% 7.608 1.765 0.299 0.047 10 2 6.98% 7.802 1.448 0.314 0.066 11 3 15.10% 8.534 1.906 0.276 0.033 11 4 42.54% 7.546 1.857 0.292 0.028

Panel B: Cultural Environment: Individualism (IDV) and Uncertainty Avoidance (UAI) Individualism (IDV)

Countries Quartile EquityPct Size MVBV LEV FCF 11 1 4.16% 8.478 1.383 0.331 0.076 10 2 2.84% 8.414 1.522 0.322 0.078 11 3 17.35% 7.045 1.946 0.256 0.019 11 4 42.27% 7.550 1.860 0.292 0.028

Uncertainty Avoidance (UAI) Countries Quartile EquityPct Size MVBV LEV FCF

11 1 42.42% 7.538 1.861 0.291 0.028 10 2 15.31% 8.453 1.781 0.316 0.037 11 3 5.57% 8.054 1.419 0.321 0.080 11 4 4.76% 8.360 1.654 0.311 0.055

50

Table IV Regression Results Using Firm-Specific and Legal Determinants of Compensation Structure

This table presents regression results from Tobit models. Our dependent variable is EquityPct, the ratio of total equity-based compensation (stock option compensation plus restricted stock compensation) to total compensation. Total compensation is the sum of cash compensation, stock option compensation, restricted stock, and long-term incentive compensation. Cash compensation is salary plus bonus. Independent variables are as follows. SelfDeal is a measure of shareholder rights protection (from Djankov et al. (2008)), where higher values suggest stronger protection of shareholder rights. Rule of Law (from La Porta et al. (1997a)) is a measure of how effectively a nation enforces its laws. Higher values indicate a stronger tradition of enforcement. Size is the natural logarithm of the firm’s total assets. Market-to-Book (MVBV) is the book value of total assets less the book value of equity plus the market value of equity divided by the book value of total assets. Leverage (LEV) is the ratio of the firm’s total debt to total assets. Free Cash Flow (FCF) is the ratio of operating income before depreciation less the sum of income tax, interest, and dividends paid scaled by the firm’s market value of equity and total debt. For all models, we include industry and year controls, but do not report coefficient values. Standard errors robust to serial correlation and heteroskedasticity are in parentheses. ***, **, * indicate significance at .01, .05, and .1, respectively.

All Firms Non-U.S. Firms

1 2 3 4

Observations 27,402 27,445 2,758 2,801

Noncensored 20,301 20,199 510 511

Intercept -0.8356*** (0.0300)

-1.8864*** (0.0443)

-1.4389*** (0.1570)

-1.4926*** (0.1717)

SelfDeal 1.1947*** (0.0402)

0.7828*** (0.1158)

Rule of Law 0.3671*** (0.0083)

0.1693*** (0.0309)

Size 0.0502*** (0.0018)

0.0530*** (0.0018)

0.0412** (0.0144)

0.0077 (0.0143)

MVBV 0.0434*** (0.0026)

0.0409*** (0.0025)

0.0993** (0.0269)

0.0955** (0.0267)

LEV -0.0951*** (0.0180)

-0.1067*** (0.0176)

-0.6724** (0.1953)

-0.4386** (0.1932)

FCF 0.0048 (0.0041)

0.0042 (0.0034)

-0.991** (0.3381)

-0.6371** (0.2461)

Industry Fixed Effects Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes

51

Table V Regression Results Using Cultural Determinants of Compensation Structure

This table presents regression results from Tobit models. Our dependent variable is EquityPct, the ratio of total equity-based compensation (stock option compensation plus restricted stock compensation) to total compensation. Total compensation is the sum of cash compensation, stock option compensation, restricted stock, and long-term incentive compensation. Cash compensation is salary plus bonus. Independent variables are as follows. IDV (from Hofstede (1980)) reflects the relation between the individual and the group. Higher scores indicate a culture in which individual interests are more likely to prevail over collective interests. UAI (from Hofstede (1980)) measures the degree to which people feel stress or discomfort in unstructured or risky circumstances. Higher values indicate less tolerance of risk. Size is the natural logarithm of the firm’s total assets. Market-to-Book (MVBV) is the book value of total assets less the book value of equity plus the market value of equity divided by the book value of total assets. Leverage (LEV) is the ratio of the firm’s total debt to total assets. Free Cash Flow (FCF) is the ratio of operating income before depreciation less the sum of income tax, interest, and dividends paid, scaled by the firm’s market value of equity and total debt. Results in Panel A include all firms (U.S. and Non-U.S.) in the sample. Panel B presents results for Non-U.S. firms only. In both panels, models 2, 3, 5, and 6 include industry and year controls (coefficients not reported), but not models 1 and 4. Standard errors robust to serial correlation and heteroskedasticity are in parentheses. ***, **, * indicate significance at .01, .05, and .1, respectively.

Panel A: All Firms

IDV UAI

Model 1 2 3 4 5 6

Observations 27,370 27,370 27,370 27,370 27,370 27,370

Noncensored 20,312 20,312 20,312 20,312 20,312 20,312

Intercept -1.1597*** (0.0357)

-1.3021*** (0.0386)

-1.6802*** (0.0396)

1.0629*** (0.0195)

0.9445*** (0.0237)

0.6591*** (0.0234)

IDV 0.0169*** (0.0004)

0.0172*** (0.0004)

0.0177*** (0.0004)

UAI -0.0152*** (0.0004)

-0.0149*** (0.0004)

-0.0157*** (0.0004)

Size 0.0581*** (0.0018)

0.0523*** (0.0018)

MVBV 0.0387*** (0.0025)

0.0421*** (0.0026)

LEV -0.1293*** (0.0174)

-0.1005*** (0.0178)

FCF 0.0042 (0.0037)

0.0045 (0.0038)

Industry Fixed Effects No Yes Yes No Yes Yes

Year Fixed Effects No Yes Yes No Yes Yes

52

Table V (continued)

Panel B: Non-U.S. Firms

IDV UAI

Model 1 2 3 4 5 6

Observations 2,726 2,726 2,726 2,726 2,726 2,726

Noncensored 518 518 518 518 518 518

Intercept -1.7124*** (0.1144)

-2.075*** (0.2289)

-1.772*** (0.1651)

-0.137* (0.0794)

-0.4321** (0.2181)

-0.5213** (0.1617)

IDV 0.0166*** (0.0014)

0.0153*** (0.0014)

0.0150*** (0.0015)

UAI -0.0100*** (0.0013)

-0.0089*** (0.0013)

-0.0082*** (0.0013)

Size 0.0274* (0.0142)

0.0369** (0.0147)

MVBV 0.0630** (0.0270)

0.1022** (0.0274)

LEV -0.5867** (0.1955)

-0.6005** (0.1984)

FCF -0.4826 (0.3357)

-1.1083** (0.3425)

Industry Fixed Effects

No Yes Yes No Yes Yes

Year Fixed Effects

No Yes Yes No Yes Yes

53

Table VI Correlation Matrix

This table presents correlations between the main legal and cultural variables that may affect compensation structure. SelfDeal is a measure of shareholder rights protection (from Djankov et al. (2008)), where higher values suggest stronger protection of shareholder rights; Rule of Law (from La Porta et al. (1997a)) is a measure of how effectively a nation enforces its laws. Higher values indicate a stronger tradition of enforcement. IDV (from Hofstede (1980)) reflects the relation between the individual and the group. UAI (from Hofstede (1980)) measures the degree to which people feel stress or discomfort in unstructured or risky circumstances. Harmony (from Schwartz (2006)) measures whether the culture espouses passive acceptance of the world as it is vs. active efforts to control the social environment through self-assertion. Catholic (from Guiso et al. (2003)) measures the percentage of the nation’s population reporting Catholicism as their religion. DistrustComp (from Aghion et al. (2010)) measures the degree of confidence in major companies as expressed by respondents to the World Values Survey (1981-2000). LangEng (from Stulz and Williamson (2003)) identifies whether English is the primary language spoken by the largest percentage of the nation’s population.

Rule of Law IDV UAI Harmony Catholic DistrustComp LangEng

SelfDeal 0.3051 0.52462 -0.6769 -0.6795 -0.5711 -0.6907 0.69544

<.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001

Rule of Law 1 0.87404 -0.5473 -0.4677 -0.1645 -0.6858 0.76273

<.0001 <.0001 <.0001 <.0001 <.0001 <.0001

IDV 1 -0.6822 -0.651 -0.3226 -0.7168 0.89993

<.0001 <.0001 <.0001 <.0001 <.0001

UAI 1 0.66134 0.48445 0.79129 -0.7537

<.0001 <.0001 <.0001 <.0001

Harmony 1 0.55685 0.69487 -0.7729

<.0001 <.0001 <.0001

Catholic 1 0.52473 -0.329

<.0001 <.0001

DistrustComp 1 -0.7671

<.0001

LangEng 1

54

Table VII

Regression Results Using Cultural and Legal Determinants of Compensation Structure

This table presents regression results from Tobit models that combine both cultural and legal determinants of compensation structure (along with firm-specific control variables). Our dependent variable is EquityPct, the ratio of total equity-based compensation to total compensation. Independent variables are as follows. IDV (from Hofstede (1980)) reflects the relation between the individual and the group. Higher scores indicate a culture in which individual interests are more likely to prevail over collective interests. UAI (from Hofstede (1980)) measures the degree to which people feel stress or discomfort in unstructured or risky circumstances. Higher values indicate less tolerance of risk. SelfDeal is a measure of shareholder rights protection (from Djankov et al. (2008)), where higher values suggest stronger protection of shareholder rights. Rule of Law (from La Porta et al. (1997a)) is a measure of how effectively a nation enforces its laws. Higher values indicate a stronger tradition of enforcement. Size is the natural logarithm of the firm’s total assets. Market-to-Book (MVBV) is the book value of total assets less the book value of equity plus the market value of equity divided by the book value of total assets. Leverage (LEV) is the ratio of the firm’s total debt to total assets. Free Cash Flow (FCF) is the ratio of operating income before depreciation less the sum of income tax, interest, and dividends paid, scaled by the firm’s market value of equity and total debt. All models include industry and year controls (coefficients not reported). Standard errors robust to serial correlation and heteroskedasticity are in parentheses. ***, **, * indicate significance at .01, .05, and .1, respectively.

55

All Firms Non-U.S. Firms

Model 1 2 3 4 5 6 7 8

Observations 27,402 27,370 27,370 27,370 2,758 2,726 2,726 2,726

Noncensored 20,301 20,287 20,287 20,287 510 507 507 507

Intercept -2.2865*** (0.0529)

-1.8871*** (0.0561)

-1.8171*** (0.0932)

-1.683*** (0.101)

-2.0792*** (0.1918)

-1.888*** (0.1858)

-1.7757*** (0.255)

-1.6901*** (0.2579)

IDV 0.0126*** (0.0008)

0.0123*** (0.0008)

0.0134*** (0.002)

0.0133*** (0.002)

UAI -0.0037*** (0.0006)

-0.0017** (0.0007)

-0.0027* (0.0016)

-0.0017 (0.0016)

SelfDeal 0.8803*** (0.0459)

0.3056*** (0.0588)

0.6373*** (0.0594)

0.2235** (0.0679)

0.8979*** (0.1167)

0.4009** (0.1315)

0.7409*** (0.1428)

0.315** (0.1535)

Rule of Law 0.3306*** (0.0085)

0.0936*** (0.0167)

0.3018*** (0.0096)

0.0852*** (0.0171)

0.2135*** (0.031)

0.0256 (0.0399)

0.2025*** (0.0317)

0.0175 (0.0406)

Size 0.0564*** (0.0017)

0.0583*** (0.0017)

0.0568*** (0.0017)

0.0584*** (0.0017)

0.0227 (0.0141)

0.0259* (0.0139)

0.0221 (0.0141)

0.0259* (0.0139)

MVBV 0.039*** (0.0025)

0.0382*** (0.0025)

0.0388*** (0.0025)

0.038*** (0.0025)

0.0683** (0.0268)

0.0432 (0.0266)

0.065** (0.0269)

0.0411 (0.0266)

LEV -0.1261*** (0.0174)

-0.1314*** (0.0174)

-0.1266*** (0.0174)

-0.1318*** (0.0174)

-0.6393** (0.1934)

-0.6247** (0.1914)

-0.6259** (0.1933)

-0.6249** (0.1912)

FCF 0.0042 (0.0036)

0.0041 (0.0037)

0.0043 (0.0036)

0.0040 (0.0037)

-0.5963* (0.3338)

-0.2388 (0.3295)

-0.5627* (0.3339)

-0.2346 (0.3292)

Industry Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes

56

Table VIII

Regression Results Using Cultural Determinants of Compensation Structure: Effect of Type of Product (Tradable vs. Non-Tradable)

This table presents regression results from Tobit models. Our dependent variable is EquityPct, the ratio of total equity-based compensation (stock option compensation plus restricted stock compensation) to total compensation. Total compensation is the sum of cash compensation, stock option compensation, restricted stock, and long-term incentive compensation. Cash compensation is salary plus bonus. Independent variables are as follows. Nontrade is an indicator variable that takes on a value of 1 if the firm produces goods that are likely to be traded only in the firms’ domestic market and 0 if the firm produces goods that are likely to trade internationally. This variable is specified by Sarkissian and Schill (2004). IDV (from Hofstede (1980)) reflects the relation between the individual and the group. Higher scores indicate a culture in which individual interests are more likely to prevail over collective interests. UAI (from Hofstede (1980)) measures the degree to which people feel stress or discomfort in unstructured or risky circumstances. Higher values indicate less tolerance of risk. Size is the natural logarithm of the firm’s total assets. Market-to-Book (MVBV) is the book value of total assets less the book value of equity plus the market value of equity divided by the book value of total assets. Leverage (LEV) is the ratio of the firm’s total debt to total assets. Free Cash Flow (FCF) is the ratio of operating income before depreciation less the sum of income tax, interest, and dividends paid, scaled by the firm’s market value of equity and total debt. Results in columns 1-3 are for all firms and results in columns 4-6 are for only Non-U.S. firms. In all models, we include year controls (coefficients not reported). Standard errors robust to serial correlation and heteroskedasticity are in parentheses. ***, **, * indicate significance at .01, .05, and .1, respectively.

57

All Firms Non-U.S. Firms

Model 1 2 3 4 5 6

Observations 27,370 27,370 27,370 2,726 2,726 2,726

Noncensored 20,287 20,287 20,287 507 507 507

Intercept -2.1818*** (0.0876)

-2.1597*** (0.1375)

-1.9652*** (0.1443)

-1.7776*** (0.2705)

-1.4972*** (0.3708)

-1.4085*** (0.3687)

Nontrade 0.5719*** (0.1135)

0.5907*** (0.1868)

0.4226** (0.2037)

-0.0888 (0.3424)

-0.8065 (0.4993)

-0.7621 (0.5055)

Nontrade*IDV 0.0103*** (0.0019)

0.0102*** (0.0019)

0.0119*** (0.0043)

0.0123*** (0.0043)

Nontrade*UAI -0.0021 (0.0013)

0.0013 (0.0014)

0.0049 (0.0032)

0.0059* (0.0033)

Nontrade*SelfDeal -0.5412*** (0.1253)

-0.1288 (0.1186)

-0.4693*** (0.1408)

0.1867 (0.2801)

1.1157*** (0.2966)

0.4900** (0.3274)

Nontrade*Rule of Law -0.2440** (0.0359)

-0.0934*** (0.0194)

-0.2339*** (0.0365)

-0.2853*** (0.0832)

-0.1087* (0.0635)

-0.2569*** (0.0842)

Nontrade*Size 0.0051 (0.0035)

0.0046 (0.0035)

0.0050 (0.0035)

0.0684** (0.0282)

0.0674** (0.0289)

0.0668** (0.0284)

Nontrade*MVBV -0.0089* (0.0050)

-0.0085* (0.0050)

-0.0087* (0.0050)

0.0160 (0.0535)

0.0406 (0.0543)

0.0255 (0.0537)

Nontrade*LEV -0.0536 (0.0348)

-0.0607*** (0.0349)

-0.0520 (0.0348)

-1.1893*** (0.3857)

-1.4592*** (0.3939)

-1.1678*** (0.3886)

Nontrade*FCF 0.0061 (0.0292)

0.0053 (0.0247)

0.0064 (0.0300)

-0.1109 (0.6685)

-0.0567 (0.6803)

-0.0791 (0.6708)

IDV 0.0090*** (0.0010)

0.0089*** (0.0010)

0.0073*** (0.0024)

0.0072*** (0.0024)

UAI -0.0025*** (0.0009)

-0.0019* (0.0010)

-0.0035 (0.0023)

-0.0032 (0.0023)

SelfDeal 0.5174*** (0.0754)

0.7303*** (0.0818)

0.4235*** (0.0906)

0.2892* (0.1725)

0.3047 (0.2011)

0.1248 (0.2075)

Rule of Law 0.1979*** (0.0229)

0.3541*** (0.0149)

0.1864*** (0.0237)

0.1479*** (0.0551)

0.2332*** (0.0470)

0.1282** (0.0567)

Size 0.0561*** (0.0025)

0.0548*** (0.0025)

0.0563*** (0.0025)

-0.0100 (0.0192)

-0.0093 (0.0198)

-0.0062 (0.0193)

MVBV 0.0409*** (0.0036)

0.0414*** (0.0036)

0.0407*** (0.0036)

0.0337 (0.0408)

0.0420 (0.0417)

0.0260 (0.0411)

LEV -0.1024*** (0.0255)

-0.0944*** (0.0256)

-0.1039*** (0.0255)

0.1021 (0.2819)

0.1518 (0.2876)

0.0455 (0.2840)

FCF 0.0033 (0.0037)

0.0033 (0.0037)

0.0033 (0.0037)

-0.0855 (0.5059)

-0.4341 (0.5099)

-0.1757 (0.5064)

Industry Fixed Effects No No No No No No

Year Fixed Effects Yes Yes Yes Yes Yes Yes

58

Table IX

Regression Results Using Cultural and Legal Factors to Explain Country-Fixed Effects

This table presents regression results from OLS models that consider how cultural and legal factors affect country fixed effects (which are first derived from models of compensation structure employing firm-level explanatory variables). In the first stage, we estimate a Tobit model where the dependent variable is the relative use of equity-based pay (EquityPct) and the independent variables are the enterprise-specific factors examined in Tables IV, V, and VII (firm size, market-to-book ratio, leverage, and free cash flow). In this first stage, we include firm fixed effects and country fixed effects to capture any unobservable time-invariant firm-level and country-level heterogeneities. The coefficient estimates on the 43 country fixed effects then become the variable of interest in the second stage of the analysis. In the results presented below, our dependent variable is the coefficient on each of the country fixed effects. Independent variables are as follows. IDV (from Hofstede (1980)) reflects the relation between the individual and the group. Higher scores indicate a culture in which individual interests are more likely to prevail over collective interests. UAI (from Hofstede (1980)) measures the degree to which people feel stress or discomfort in unstructured or risky circumstances. Higher values indicate less tolerance of risk. SelfDeal is a measure of shareholder rights protection (from Djankov et al. (2008)), where higher values suggest stronger protection of shareholder rights. Rule of Law (from La Porta et al. (1997a)) is a measure of how effectively a nation enforces its laws. Higher values indicate a stronger tradition of enforcement. Standard errors are in parentheses. ***, **, * indicate significance at .01, .05, and .1, respectively.

Model 1 2 3 4 5 6 7

Intercept 58.920*** (4.023)

59.017*** (4.521)

0.527*** (0.043)

3.899*** (0.200)

-1.232 (0.862)

-2.100 (1.140)

-1.967 (1.271)

IDV 8.466*** (2.374)

0.0273*** (0.009)

0.027*** (0.009)

0.029** (0.012)

UAI -6.762** (2.668)

-0.016* (0.009)

-0.011 (0.010)

-0.011 (0.010)

SelfDeal 0.053** (0.026)

1.173 (1.017)

1.128 (1.048)

Rule of Law 0.252** (0.124)

-0.057 (0.230)

Adj R2 0.2509 0.1342 0.0817 0.0707 0.2982 0.3052 0.2842

59

Table X

Robustness Testing: Regression Results Using Alternative Instruments of Culture

This table presents regression results from Tobit models that include alternative instruments of cultural factors (along with the legal variables and the firm-specific control variables). Our dependent variable is EquityPct, the ratio of total equity-based compensation to total compensation. Independent variables are as follows. Harmony (from Schwartz (2006)) measures whether the culture espouses passive acceptance of the world as it is vs. active efforts to control the social environment through self-assertion. Catholic (from Guiso et al. (2003)) measures the percentage of the nation’s population reporting Catholicism as their religion. LangEng (from Stulz and Williamson (2003)) identifies whether English is the primary language spoken by the largest percentage of the nation’s population. DistrustComp (from Aghion et al. (2010)) measures the degree of confidence in major companies as expressed by respondents to the World Values Survey (1981-2000). Higher scores indicate less confidence. SelfDeal is a measure of shareholder rights protection (from Djankov et al. (2008)), where higher values suggest stronger protection of shareholder rights. Rule of Law (from La Porta et al. (1997a)) is a measure of how effectively a nation enforces its laws. Higher values indicate a stronger tradition of enforcement. The firm-specific variables are as follows. Size is the natural logarithm of the firm’s total assets. Market-to-Book (MVBV) is the book value of total assets less the book value of equity plus the market value of equity divided by the book value of total assets. Leverage (LEV) is the ratio of the firm’s total debt to total assets. Free Cash Flow (FCF) is the ratio of operating income before depreciation less the sum of income tax, interest, and dividends paid, scaled by the firm’s market value of equity and total debt. All models include industry and year controls (coefficients not reported). Standard errors robust to serial correlation and heteroskedasticity are in parentheses. ***, **, * indicate significance at .01, .05, and .1, respectively. Panel A presents results for all firms and Panel B presents results for Non-U.S. firms only.

60

Table X (continued) Panel A: All Firms

1 2 3 4

Observations 27,347 27,445 27,445 27,325

Noncensored 20,213 20,241 20,241 20,258

Intercept 0.7108*** (0.1561)

-2.2185*** (0.0584)

-1.2855*** (0.0716)

-1.8573*** (0.1205)

Harmony -0.6283*** (0.0306)

Catholic -0.0011*** (0.0004)

LangEng 0.5690*** (0.0337)

DistrustComp -2.3185*** (0.3649)

SelfDeal 0.1616*** (0.056)

0.8244*** (0.0504)

-0.0796 (0.0700)

0.5319*** (0.0666)

Rule of Law 0.2552*** (0.0085)

0.3308*** (0.0085)

0.1392*** (0.0132)

0.3229*** (0.0141)

Size 0.0617*** (0.0018)

0.0562*** (0.0018)

0.0591*** (0.0017)

0.0585*** (0.0018)

MVBV 0.0380*** (0.0025)

0.0390*** (0.0025)

0.0389*** (0.0025)

0.0395*** (0.0025)

LEV -0.0914*** (0.0177)

-0.1246*** (0.0174)

-0.1387*** (0.0173)

-0.1233*** (0.0174)

FCF 0.0032 (0.003)

0.0042 (0.0036)

0.0039 (0.0033)

0.0041 (0.0036)

Industry Fixed Effects

Yes Yes Yes Yes

Year Fixed Effects

Yes Yes Yes Yes

61

Table X (continued)

Panel B: Non-U.S. Firms

1 2 3 4

Observations 2,703 2,801 2,801 2,681

Noncensored 486 513 513 425

Intercept -0.9749*** (0.3516)

-1.8936*** (0.2012)

-1.9541*** (0.2172)

-2.0949*** (0.3063)

Harmony -0.5375*** (0.1199)

Catholic -0.0026** (0.0009)

LangEng 0.1255 (0.1104)

DistrustComp -1.3452* (0.7626)

SelfDeal 0.4450*** (0.1445)

0.7561*** (0.1279)

0.7244*** (0.1901)

0.7214*** (0.1635)

Rule of Law 0.3066*** (0.0369)

0.2226*** (0.0312)

0.1970*** (0.0339)

0.2501*** (0.0420)

Size 0.0181 (0.0147)

0.0153 (0.0145)

0.0241* (0.0142)

0.0206 (0.0157)

MVBV 0.0369 (0.0275)

0.0645** (0.0269)

0.0657** (0.0268)

0.0744** (0.0302)

LEV -0.2481 (0.202)

-0.5983*** (0.1943)

-0.6576*** (0.1937)

-0.4502** (0.2202)

FCF -0.4148 (0.3285)

-0.5517* (0.3345)

-0.5837* (0.3326)

-0.5570 (0.3773)

Industry Fixed Effects

Yes Yes Yes Yes

Year Fixed Effects

Yes Yes Yes Yes

This figuof total cdata are a

ure presents compensationarrayed base

the average n) per year f

ed on nationa

F

value of Eqfrom 1996-2ality: U.S. an

62

Figure 1

quityPct (equ2009 for our nd Non-U.S.

uity-based csample of f

. firms.

compensationfirms from 4

n as a percen43 countries.

ntage The

This figuof total cdata are a(English non-U.S.

ure presents compensationarrayed by ncommon la

. firms (non-

the average n) per year f

nationality anaw legal sys-English com

F

value of Eqfrom 1996-2nd by origin tem), non-U

mmon law le

63

Figure 2

quityPct (equ2009 for our

of legal systU.S. firms (Egal system).

uity-based csample of f

tem. Our thEnglish com.

compensationfirms from 4hree groupingmmon law le

n as a percen43 countries.gs are: U.S. egal system)

ntage The firms

), and

This figuof total cbased on(1980), ththe groupprevail o

ure presents compensationn each counhe IDV scorp. Higher IDver collectiv

the average n). The data

ntry’s score re reflects hoDV scores inve interests.

F

value of Eqa are from aon the Indi

ow each natndicate a cu

64

Figure 3

quityPct (equall nations inividualism (ion specifieslture in whic

uity-based cn our sample(IDV) metris the relationch individua

compensatione and are arric. Developn between thal interests a

n as a percenrayed by quaped by Hofhe individuaare more like

ntage artiles fstede al and ely to

This figuof total cbased onHofstedediscomfouncertain

ure presents compensationn each coune (1980), theort in unstrucnty.

the average n). The datantry’s score e UAI measuctured or ris

F

value of Eqa are from aon the Un

ures the degrsky circumst

65

Figure 4

quityPct (equall nations inncertainty Aree to whichtances. High

uity-based cn our sample

Avoidance Inh people fromher UAI valu

compensatione and are arrndex (UAI).m each counues indicate

n as a percenrayed by qua. Developentry feel stre less toleran

ntage artiles ed by ess or nce of

66

Figure 5

This figure presents the average value of EquityPct (equity-based compensation as a percentage of total compensation). The data are from the non-U.S. nations in our sample and are arrayed by quartiles based on each country’s score on the Individualism (IDV) metric. Developed by Hofstede (1980), the IDV score reflects how each nation specifies the relation between the individual and the group. Higher IDV scores indicate a culture in which individual interests are more likely to prevail over collective interests.

0%

5%

10%

15%

20%

25%

1 2 3 4

EquityPct by IDV Quartile (Non US Firms)

67

Figure 6

This figure presents the average value of EquityPct (equity-based compensation as a percentage of total compensation). The data are from the non-U.S. nations in our sample and are arrayed by quartiles based on each country’s score on the Uncertainty Avoidance Index (UAI). Developed by Hofstede (1980), the UAI measures the degree to which people from each country feel stress or discomfort in unstructured or risky circumstances. Higher UAI values indicate less tolerance of uncertainty.

0%

2%

4%

6%

8%

10%

12%

14%

16%

18%

1 2 3 4

EquityPct by UAI Quartile (Non-US Firms)